/[pcre]/code/trunk/ChangeLog
ViewVC logotype

Contents of /code/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 697 - (show annotations)
Mon Sep 19 12:28:24 2011 UTC (3 years, 9 months ago) by ph10
File size: 214114 byte(s)
Error occurred while calculating annotation data.
Fix minimum length bug with *ACCEPT.
1 ChangeLog for PCRE
2 ------------------
3
4 Version 8.20 12-Sep-2011
5 ------------------------
6
7 1. Change 37 of 8.13 broke patterns like [:a]...[b:] because it thought it had
8 a POSIX class. After further experiments with Perl, which convinced me that
9 Perl has bugs and confusions, a closing square bracket is no longer allowed
10 in a POSIX name. This bug also affected patterns with classes that started
11 with full stops.
12
13 2. If a pattern such as /(a)b|ac/ is matched against "ac", there is no captured
14 substring, but while checking the failing first alternative, substring 1 is
15 temporarily captured. If the output vector supplied to pcre_exec() was not
16 big enough for this capture, the yield of the function was still zero
17 ("insufficient space for captured substrings"). This cannot be totally fixed
18 without adding another stack variable, which seems a lot of expense for a
19 edge case. However, I have improved the situation in cases such as
20 /(a)(b)x|abc/ matched against "abc", where the return code indicates that
21 fewer than the maximum number of slots in the ovector have been set.
22
23 3. Related to (2) above: when there are more back references in a pattern than
24 slots in the output vector, pcre_exec() uses temporary memory during
25 matching, and copies in the captures as far as possible afterwards. It was
26 using the entire output vector, but this conflicts with the specification
27 that only 2/3 is used for passing back captured substrings. Now it uses only
28 the first 2/3, for compatibility. This is, of course, another edge case.
29
30 4. Zoltan Herczeg's just-in-time compiler support has been integrated into the
31 main code base, and can be used by building with --enable-jit. When this is
32 done, pcregrep automatically uses it unless --disable-pcregrep-jit or the
33 runtime --no-jit option is given.
34
35 5. When the number of matches in a pcre_dfa_exec() run exactly filled the
36 ovector, the return from the function was zero, implying that there were
37 other matches that did not fit. The correct "exactly full" value is now
38 returned.
39
40 6. If a subpattern that was called recursively or as a subroutine contained
41 (*PRUNE) or any other control that caused it to give a non-standard return,
42 invalid errors such as "Error -26 (nested recursion at the same subject
43 position)" or even infinite loops could occur.
44
45 7. If a pattern such as /a(*SKIP)c|b(*ACCEPT)|/ was studied, it stopped
46 computing the minimum length on reaching *ACCEPT, and so ended up with the
47 wrong value of 1 rather than 0. Further investigation indicates that
48 computing a minimum subject length in the presence of *ACCEPT is difficult
49 (think back references, subroutine calls), and so I have changed the code so
50 that no minimum is registered for a pattern that contains *ACCEPT.
51
52
53 Version 8.13 16-Aug-2011
54 ------------------------
55
56 1. The Unicode data tables have been updated to Unicode 6.0.0.
57
58 2. Two minor typos in pcre_internal.h have been fixed.
59
60 3. Added #include <string.h> to pcre_scanner_unittest.cc, pcrecpp.cc, and
61 pcrecpp_unittest.cc. They are needed for strcmp(), memset(), and strchr()
62 in some environments (e.g. Solaris 10/SPARC using Sun Studio 12U2).
63
64 4. There were a number of related bugs in the code for matching backrefences
65 caselessly in UTF-8 mode when codes for the characters concerned were
66 different numbers of bytes. For example, U+023A and U+2C65 are an upper
67 and lower case pair, using 2 and 3 bytes, respectively. The main bugs were:
68 (a) A reference to 3 copies of a 2-byte code matched only 2 of a 3-byte
69 code. (b) A reference to 2 copies of a 3-byte code would not match 2 of a
70 2-byte code at the end of the subject (it thought there wasn't enough data
71 left).
72
73 5. Comprehensive information about what went wrong is now returned by
74 pcre_exec() and pcre_dfa_exec() when the UTF-8 string check fails, as long
75 as the output vector has at least 2 elements. The offset of the start of
76 the failing character and a reason code are placed in the vector.
77
78 6. When the UTF-8 string check fails for pcre_compile(), the offset that is
79 now returned is for the first byte of the failing character, instead of the
80 last byte inspected. This is an incompatible change, but I hope it is small
81 enough not to be a problem. It makes the returned offset consistent with
82 pcre_exec() and pcre_dfa_exec().
83
84 7. pcretest now gives a text phrase as well as the error number when
85 pcre_exec() or pcre_dfa_exec() fails; if the error is a UTF-8 check
86 failure, the offset and reason code are output.
87
88 8. When \R was used with a maximizing quantifier it failed to skip backwards
89 over a \r\n pair if the subsequent match failed. Instead, it just skipped
90 back over a single character (\n). This seems wrong (because it treated the
91 two characters as a single entity when going forwards), conflicts with the
92 documentation that \R is equivalent to (?>\r\n|\n|...etc), and makes the
93 behaviour of \R* different to (\R)*, which also seems wrong. The behaviour
94 has been changed.
95
96 9. Some internal refactoring has changed the processing so that the handling
97 of the PCRE_CASELESS and PCRE_MULTILINE options is done entirely at compile
98 time (the PCRE_DOTALL option was changed this way some time ago: version
99 7.7 change 16). This has made it possible to abolish the OP_OPT op code,
100 which was always a bit of a fudge. It also means that there is one less
101 argument for the match() function, which reduces its stack requirements
102 slightly. This change also fixes an incompatibility with Perl: the pattern
103 (?i:([^b]))(?1) should not match "ab", but previously PCRE gave a match.
104
105 10. More internal refactoring has drastically reduced the number of recursive
106 calls to match() for possessively repeated groups such as (abc)++ when
107 using pcre_exec().
108
109 11. While implementing 10, a number of bugs in the handling of groups were
110 discovered and fixed:
111
112 (?<=(a)+) was not diagnosed as invalid (non-fixed-length lookbehind).
113 (a|)*(?1) gave a compile-time internal error.
114 ((a|)+)+ did not notice that the outer group could match an empty string.
115 (^a|^)+ was not marked as anchored.
116 (.*a|.*)+ was not marked as matching at start or after a newline.
117
118 12. Yet more internal refactoring has removed another argument from the match()
119 function. Special calls to this function are now indicated by setting a
120 value in a variable in the "match data" data block.
121
122 13. Be more explicit in pcre_study() instead of relying on "default" for
123 opcodes that mean there is no starting character; this means that when new
124 ones are added and accidentally left out of pcre_study(), testing should
125 pick them up.
126
127 14. The -s option of pcretest has been documented for ages as being an old
128 synonym of -m (show memory usage). I have changed it to mean "force study
129 for every regex", that is, assume /S for every regex. This is similar to -i
130 and -d etc. It's slightly incompatible, but I'm hoping nobody is still
131 using it. It makes it easier to run collections of tests with and without
132 study enabled, and thereby test pcre_study() more easily. All the standard
133 tests are now run with and without -s (but some patterns can be marked as
134 "never study" - see 20 below).
135
136 15. When (*ACCEPT) was used in a subpattern that was called recursively, the
137 restoration of the capturing data to the outer values was not happening
138 correctly.
139
140 16. If a recursively called subpattern ended with (*ACCEPT) and matched an
141 empty string, and PCRE_NOTEMPTY was set, pcre_exec() thought the whole
142 pattern had matched an empty string, and so incorrectly returned a no
143 match.
144
145 17. There was optimizing code for the last branch of non-capturing parentheses,
146 and also for the obeyed branch of a conditional subexpression, which used
147 tail recursion to cut down on stack usage. Unfortunately, now that there is
148 the possibility of (*THEN) occurring in these branches, tail recursion is
149 no longer possible because the return has to be checked for (*THEN). These
150 two optimizations have therefore been removed.
151
152 18. If a pattern containing \R was studied, it was assumed that \R always
153 matched two bytes, thus causing the minimum subject length to be
154 incorrectly computed because \R can also match just one byte.
155
156 19. If a pattern containing (*ACCEPT) was studied, the minimum subject length
157 was incorrectly computed.
158
159 20. If /S is present twice on a test pattern in pcretest input, it now
160 *disables* studying, thereby overriding the use of -s on the command line
161 (see 14 above). This is necessary for one or two tests to keep the output
162 identical in both cases.
163
164 21. When (*ACCEPT) was used in an assertion that matched an empty string and
165 PCRE_NOTEMPTY was set, PCRE applied the non-empty test to the assertion.
166
167 22. When an atomic group that contained a capturing parenthesis was
168 successfully matched, but the branch in which it appeared failed, the
169 capturing was not being forgotten if a higher numbered group was later
170 captured. For example, /(?>(a))b|(a)c/ when matching "ac" set capturing
171 group 1 to "a", when in fact it should be unset. This applied to multi-
172 branched capturing and non-capturing groups, repeated or not, and also to
173 positive assertions (capturing in negative assertions does not happen
174 in PCRE) and also to nested atomic groups.
175
176 23. Add the ++ qualifier feature to pcretest, to show the remainder of the
177 subject after a captured substring, to make it easier to tell which of a
178 number of identical substrings has been captured.
179
180 24. The way atomic groups are processed by pcre_exec() has been changed so that
181 if they are repeated, backtracking one repetition now resets captured
182 values correctly. For example, if ((?>(a+)b)+aabab) is matched against
183 "aaaabaaabaabab" the value of captured group 2 is now correctly recorded as
184 "aaa". Previously, it would have been "a". As part of this code
185 refactoring, the way recursive calls are handled has also been changed.
186
187 25. If an assertion condition captured any substrings, they were not passed
188 back unless some other capturing happened later. For example, if
189 (?(?=(a))a) was matched against "a", no capturing was returned.
190
191 26. When studying a pattern that contained subroutine calls or assertions,
192 the code for finding the minimum length of a possible match was handling
193 direct recursions such as (xxx(?1)|yyy) but not mutual recursions (where
194 group 1 called group 2 while simultaneously a separate group 2 called group
195 1). A stack overflow occurred in this case. I have fixed this by limiting
196 the recursion depth to 10.
197
198 27. Updated RunTest.bat in the distribution to the version supplied by Tom
199 Fortmann. This supports explicit test numbers on the command line, and has
200 argument validation and error reporting.
201
202 28. An instance of \X with an unlimited repeat could fail if at any point the
203 first character it looked at was a mark character.
204
205 29. Some minor code refactoring concerning Unicode properties and scripts
206 should reduce the stack requirement of match() slightly.
207
208 30. Added the '=' option to pcretest to check the setting of unused capturing
209 slots at the end of the pattern, which are documented as being -1, but are
210 not included in the return count.
211
212 31. If \k was not followed by a braced, angle-bracketed, or quoted name, PCRE
213 compiled something random. Now it gives a compile-time error (as does
214 Perl).
215
216 32. A *MARK encountered during the processing of a positive assertion is now
217 recorded and passed back (compatible with Perl).
218
219 33. If --only-matching or --colour was set on a pcregrep call whose pattern
220 had alternative anchored branches, the search for a second match in a line
221 was done as if at the line start. Thus, for example, /^01|^02/ incorrectly
222 matched the line "0102" twice. The same bug affected patterns that started
223 with a backwards assertion. For example /\b01|\b02/ also matched "0102"
224 twice.
225
226 34. Previously, PCRE did not allow quantification of assertions. However, Perl
227 does, and because of capturing effects, quantifying parenthesized
228 assertions may at times be useful. Quantifiers are now allowed for
229 parenthesized assertions.
230
231 35. A minor code tidy in pcre_compile() when checking options for \R usage.
232
233 36. \g was being checked for fancy things in a character class, when it should
234 just be a literal "g".
235
236 37. PCRE was rejecting [:a[:digit:]] whereas Perl was not. It seems that the
237 appearance of a nested POSIX class supersedes an apparent external class.
238 For example, [:a[:digit:]b:] matches "a", "b", ":", or a digit. Also,
239 unescaped square brackets may also appear as part of class names. For
240 example, [:a[:abc]b:] gives unknown class "[:abc]b:]". PCRE now behaves
241 more like Perl. (But see 8.20/1 above.)
242
243 38. PCRE was giving an error for \N with a braced quantifier such as {1,} (this
244 was because it thought it was \N{name}, which is not supported).
245
246 39. Add minix to OS list not supporting the -S option in pcretest.
247
248 40. PCRE tries to detect cases of infinite recursion at compile time, but it
249 cannot analyze patterns in sufficient detail to catch mutual recursions
250 such as ((?1))((?2)). There is now a runtime test that gives an error if a
251 subgroup is called recursively as a subpattern for a second time at the
252 same position in the subject string. In previous releases this might have
253 been caught by the recursion limit, or it might have run out of stack.
254
255 41. A pattern such as /(?(R)a+|(?R)b)/ is quite safe, as the recursion can
256 happen only once. PCRE was, however incorrectly giving a compile time error
257 "recursive call could loop indefinitely" because it cannot analyze the
258 pattern in sufficient detail. The compile time test no longer happens when
259 PCRE is compiling a conditional subpattern, but actual runaway loops are
260 now caught at runtime (see 40 above).
261
262 42. It seems that Perl allows any characters other than a closing parenthesis
263 to be part of the NAME in (*MARK:NAME) and other backtracking verbs. PCRE
264 has been changed to be the same.
265
266 43. Updated configure.ac to put in more quoting round AC_LANG_PROGRAM etc. so
267 as not to get warnings when autogen.sh is called. Also changed
268 AC_PROG_LIBTOOL (deprecated) to LT_INIT (the current macro).
269
270 44. To help people who use pcregrep to scan files containing exceedingly long
271 lines, the following changes have been made:
272
273 (a) The default value of the buffer size parameter has been increased from
274 8K to 20K. (The actual buffer used is three times this size.)
275
276 (b) The default can be changed by ./configure --with-pcregrep-bufsize when
277 PCRE is built.
278
279 (c) A --buffer-size=n option has been added to pcregrep, to allow the size
280 to be set at run time.
281
282 (d) Numerical values in pcregrep options can be followed by K or M, for
283 example --buffer-size=50K.
284
285 (e) If a line being scanned overflows pcregrep's buffer, an error is now
286 given and the return code is set to 2.
287
288 45. Add a pointer to the latest mark to the callout data block.
289
290 46. The pattern /.(*F)/, when applied to "abc" with PCRE_PARTIAL_HARD, gave a
291 partial match of an empty string instead of no match. This was specific to
292 the use of ".".
293
294 47. The pattern /f.*/8s, when applied to "for" with PCRE_PARTIAL_HARD, gave a
295 complete match instead of a partial match. This bug was dependent on both
296 the PCRE_UTF8 and PCRE_DOTALL options being set.
297
298 48. For a pattern such as /\babc|\bdef/ pcre_study() was failing to set up the
299 starting byte set, because \b was not being ignored.
300
301
302 Version 8.12 15-Jan-2011
303 ------------------------
304
305 1. Fixed some typos in the markup of the man pages, and wrote a script that
306 checks for such things as part of the documentation building process.
307
308 2. On a big-endian 64-bit system, pcregrep did not correctly process the
309 --match-limit and --recursion-limit options (added for 8.11). In
310 particular, this made one of the standard tests fail. (The integer value
311 went into the wrong half of a long int.)
312
313 3. If the --colour option was given to pcregrep with -v (invert match), it
314 did strange things, either producing crazy output, or crashing. It should,
315 of course, ignore a request for colour when reporting lines that do not
316 match.
317
318 4. Another pcregrep bug caused similar problems if --colour was specified with
319 -M (multiline) and the pattern match finished with a line ending.
320
321 5. In pcregrep, when a pattern that ended with a literal newline sequence was
322 matched in multiline mode, the following line was shown as part of the
323 match. This seems wrong, so I have changed it.
324
325 6. Another pcregrep bug in multiline mode, when --colour was specified, caused
326 the check for further matches in the same line (so they could be coloured)
327 to overrun the end of the current line. If another match was found, it was
328 incorrectly shown (and then shown again when found in the next line).
329
330 7. If pcregrep was compiled under Windows, there was a reference to the
331 function pcregrep_exit() before it was defined. I am assuming this was
332 the cause of the "error C2371: 'pcregrep_exit' : redefinition;" that was
333 reported by a user. I've moved the definition above the reference.
334
335
336 Version 8.11 10-Dec-2010
337 ------------------------
338
339 1. (*THEN) was not working properly if there were untried alternatives prior
340 to it in the current branch. For example, in ((a|b)(*THEN)(*F)|c..) it
341 backtracked to try for "b" instead of moving to the next alternative branch
342 at the same level (in this case, to look for "c"). The Perl documentation
343 is clear that when (*THEN) is backtracked onto, it goes to the "next
344 alternative in the innermost enclosing group".
345
346 2. (*COMMIT) was not overriding (*THEN), as it does in Perl. In a pattern
347 such as (A(*COMMIT)B(*THEN)C|D) any failure after matching A should
348 result in overall failure. Similarly, (*COMMIT) now overrides (*PRUNE) and
349 (*SKIP), (*SKIP) overrides (*PRUNE) and (*THEN), and (*PRUNE) overrides
350 (*THEN).
351
352 3. If \s appeared in a character class, it removed the VT character from
353 the class, even if it had been included by some previous item, for example
354 in [\x00-\xff\s]. (This was a bug related to the fact that VT is not part
355 of \s, but is part of the POSIX "space" class.)
356
357 4. A partial match never returns an empty string (because you can always
358 match an empty string at the end of the subject); however the checking for
359 an empty string was starting at the "start of match" point. This has been
360 changed to the "earliest inspected character" point, because the returned
361 data for a partial match starts at this character. This means that, for
362 example, /(?<=abc)def/ gives a partial match for the subject "abc"
363 (previously it gave "no match").
364
365 5. Changes have been made to the way PCRE_PARTIAL_HARD affects the matching
366 of $, \z, \Z, \b, and \B. If the match point is at the end of the string,
367 previously a full match would be given. However, setting PCRE_PARTIAL_HARD
368 has an implication that the given string is incomplete (because a partial
369 match is preferred over a full match). For this reason, these items now
370 give a partial match in this situation. [Aside: previously, the one case
371 /t\b/ matched against "cat" with PCRE_PARTIAL_HARD set did return a partial
372 match rather than a full match, which was wrong by the old rules, but is
373 now correct.]
374
375 6. There was a bug in the handling of #-introduced comments, recognized when
376 PCRE_EXTENDED is set, when PCRE_NEWLINE_ANY and PCRE_UTF8 were also set.
377 If a UTF-8 multi-byte character included the byte 0x85 (e.g. +U0445, whose
378 UTF-8 encoding is 0xd1,0x85), this was misinterpreted as a newline when
379 scanning for the end of the comment. (*Character* 0x85 is an "any" newline,
380 but *byte* 0x85 is not, in UTF-8 mode). This bug was present in several
381 places in pcre_compile().
382
383 7. Related to (6) above, when pcre_compile() was skipping #-introduced
384 comments when looking ahead for named forward references to subpatterns,
385 the only newline sequence it recognized was NL. It now handles newlines
386 according to the set newline convention.
387
388 8. SunOS4 doesn't have strerror() or strtoul(); pcregrep dealt with the
389 former, but used strtoul(), whereas pcretest avoided strtoul() but did not
390 cater for a lack of strerror(). These oversights have been fixed.
391
392 9. Added --match-limit and --recursion-limit to pcregrep.
393
394 10. Added two casts needed to build with Visual Studio when NO_RECURSE is set.
395
396 11. When the -o option was used, pcregrep was setting a return code of 1, even
397 when matches were found, and --line-buffered was not being honoured.
398
399 12. Added an optional parentheses number to the -o and --only-matching options
400 of pcregrep.
401
402 13. Imitating Perl's /g action for multiple matches is tricky when the pattern
403 can match an empty string. The code to do it in pcretest and pcredemo
404 needed fixing:
405
406 (a) When the newline convention was "crlf", pcretest got it wrong, skipping
407 only one byte after an empty string match just before CRLF (this case
408 just got forgotten; "any" and "anycrlf" were OK).
409
410 (b) The pcretest code also had a bug, causing it to loop forever in UTF-8
411 mode when an empty string match preceded an ASCII character followed by
412 a non-ASCII character. (The code for advancing by one character rather
413 than one byte was nonsense.)
414
415 (c) The pcredemo.c sample program did not have any code at all to handle
416 the cases when CRLF is a valid newline sequence.
417
418 14. Neither pcre_exec() nor pcre_dfa_exec() was checking that the value given
419 as a starting offset was within the subject string. There is now a new
420 error, PCRE_ERROR_BADOFFSET, which is returned if the starting offset is
421 negative or greater than the length of the string. In order to test this,
422 pcretest is extended to allow the setting of negative starting offsets.
423
424 15. In both pcre_exec() and pcre_dfa_exec() the code for checking that the
425 starting offset points to the beginning of a UTF-8 character was
426 unnecessarily clumsy. I tidied it up.
427
428 16. Added PCRE_ERROR_SHORTUTF8 to make it possible to distinguish between a
429 bad UTF-8 sequence and one that is incomplete when using PCRE_PARTIAL_HARD.
430
431 17. Nobody had reported that the --include_dir option, which was added in
432 release 7.7 should have been called --include-dir (hyphen, not underscore)
433 for compatibility with GNU grep. I have changed it to --include-dir, but
434 left --include_dir as an undocumented synonym, and the same for
435 --exclude-dir, though that is not available in GNU grep, at least as of
436 release 2.5.4.
437
438 18. At a user's suggestion, the macros GETCHAR and friends (which pick up UTF-8
439 characters from a string of bytes) have been redefined so as not to use
440 loops, in order to improve performance in some environments. At the same
441 time, I abstracted some of the common code into auxiliary macros to save
442 repetition (this should not affect the compiled code).
443
444 19. If \c was followed by a multibyte UTF-8 character, bad things happened. A
445 compile-time error is now given if \c is not followed by an ASCII
446 character, that is, a byte less than 128. (In EBCDIC mode, the code is
447 different, and any byte value is allowed.)
448
449 20. Recognize (*NO_START_OPT) at the start of a pattern to set the PCRE_NO_
450 START_OPTIMIZE option, which is now allowed at compile time - but just
451 passed through to pcre_exec() or pcre_dfa_exec(). This makes it available
452 to pcregrep and other applications that have no direct access to PCRE
453 options. The new /Y option in pcretest sets this option when calling
454 pcre_compile().
455
456 21. Change 18 of release 8.01 broke the use of named subpatterns for recursive
457 back references. Groups containing recursive back references were forced to
458 be atomic by that change, but in the case of named groups, the amount of
459 memory required was incorrectly computed, leading to "Failed: internal
460 error: code overflow". This has been fixed.
461
462 22. Some patches to pcre_stringpiece.h, pcre_stringpiece_unittest.cc, and
463 pcretest.c, to avoid build problems in some Borland environments.
464
465
466 Version 8.10 25-Jun-2010
467 ------------------------
468
469 1. Added support for (*MARK:ARG) and for ARG additions to PRUNE, SKIP, and
470 THEN.
471
472 2. (*ACCEPT) was not working when inside an atomic group.
473
474 3. Inside a character class, \B is treated as a literal by default, but
475 faulted if PCRE_EXTRA is set. This mimics Perl's behaviour (the -w option
476 causes the error). The code is unchanged, but I tidied the documentation.
477
478 4. Inside a character class, PCRE always treated \R and \X as literals,
479 whereas Perl faults them if its -w option is set. I have changed PCRE so
480 that it faults them when PCRE_EXTRA is set.
481
482 5. Added support for \N, which always matches any character other than
483 newline. (It is the same as "." when PCRE_DOTALL is not set.)
484
485 6. When compiling pcregrep with newer versions of gcc which may have
486 FORTIFY_SOURCE set, several warnings "ignoring return value of 'fwrite',
487 declared with attribute warn_unused_result" were given. Just casting the
488 result to (void) does not stop the warnings; a more elaborate fudge is
489 needed. I've used a macro to implement this.
490
491 7. Minor change to pcretest.c to avoid a compiler warning.
492
493 8. Added four artifical Unicode properties to help with an option to make
494 \s etc use properties (see next item). The new properties are: Xan
495 (alphanumeric), Xsp (Perl space), Xps (POSIX space), and Xwd (word).
496
497 9. Added PCRE_UCP to make \b, \d, \s, \w, and certain POSIX character classes
498 use Unicode properties. (*UCP) at the start of a pattern can be used to set
499 this option. Modified pcretest to add /W to test this facility. Added
500 REG_UCP to make it available via the POSIX interface.
501
502 10. Added --line-buffered to pcregrep.
503
504 11. In UTF-8 mode, if a pattern that was compiled with PCRE_CASELESS was
505 studied, and the match started with a letter with a code point greater than
506 127 whose first byte was different to the first byte of the other case of
507 the letter, the other case of this starting letter was not recognized
508 (#976).
509
510 12. If a pattern that was studied started with a repeated Unicode property
511 test, for example, \p{Nd}+, there was the theoretical possibility of
512 setting up an incorrect bitmap of starting bytes, but fortunately it could
513 not have actually happened in practice until change 8 above was made (it
514 added property types that matched character-matching opcodes).
515
516 13. pcre_study() now recognizes \h, \v, and \R when constructing a bit map of
517 possible starting bytes for non-anchored patterns.
518
519 14. Extended the "auto-possessify" feature of pcre_compile(). It now recognizes
520 \R, and also a number of cases that involve Unicode properties, both
521 explicit and implicit when PCRE_UCP is set.
522
523 15. If a repeated Unicode property match (e.g. \p{Lu}*) was used with non-UTF-8
524 input, it could crash or give wrong results if characters with values
525 greater than 0xc0 were present in the subject string. (Detail: it assumed
526 UTF-8 input when processing these items.)
527
528 16. Added a lot of (int) casts to avoid compiler warnings in systems where
529 size_t is 64-bit (#991).
530
531 17. Added a check for running out of memory when PCRE is compiled with
532 --disable-stack-for-recursion (#990).
533
534 18. If the last data line in a file for pcretest does not have a newline on
535 the end, a newline was missing in the output.
536
537 19. The default pcre_chartables.c file recognizes only ASCII characters (values
538 less than 128) in its various bitmaps. However, there is a facility for
539 generating tables according to the current locale when PCRE is compiled. It
540 turns out that in some environments, 0x85 and 0xa0, which are Unicode space
541 characters, are recognized by isspace() and therefore were getting set in
542 these tables, and indeed these tables seem to approximate to ISO 8859. This
543 caused a problem in UTF-8 mode when pcre_study() was used to create a list
544 of bytes that can start a match. For \s, it was including 0x85 and 0xa0,
545 which of course cannot start UTF-8 characters. I have changed the code so
546 that only real ASCII characters (less than 128) and the correct starting
547 bytes for UTF-8 encodings are set for characters greater than 127 when in
548 UTF-8 mode. (When PCRE_UCP is set - see 9 above - the code is different
549 altogether.)
550
551 20. Added the /T option to pcretest so as to be able to run tests with non-
552 standard character tables, thus making it possible to include the tests
553 used for 19 above in the standard set of tests.
554
555 21. A pattern such as (?&t)(?#()(?(DEFINE)(?<t>a)) which has a forward
556 reference to a subpattern the other side of a comment that contains an
557 opening parenthesis caused either an internal compiling error, or a
558 reference to the wrong subpattern.
559
560
561 Version 8.02 19-Mar-2010
562 ------------------------
563
564 1. The Unicode data tables have been updated to Unicode 5.2.0.
565
566 2. Added the option --libs-cpp to pcre-config, but only when C++ support is
567 configured.
568
569 3. Updated the licensing terms in the pcregexp.pas file, as agreed with the
570 original author of that file, following a query about its status.
571
572 4. On systems that do not have stdint.h (e.g. Solaris), check for and include
573 inttypes.h instead. This fixes a bug that was introduced by change 8.01/8.
574
575 5. A pattern such as (?&t)*+(?(DEFINE)(?<t>.)) which has a possessive
576 quantifier applied to a forward-referencing subroutine call, could compile
577 incorrect code or give the error "internal error: previously-checked
578 referenced subpattern not found".
579
580 6. Both MS Visual Studio and Symbian OS have problems with initializing
581 variables to point to external functions. For these systems, therefore,
582 pcre_malloc etc. are now initialized to local functions that call the
583 relevant global functions.
584
585 7. There were two entries missing in the vectors called coptable and poptable
586 in pcre_dfa_exec.c. This could lead to memory accesses outsize the vectors.
587 I've fixed the data, and added a kludgy way of testing at compile time that
588 the lengths are correct (equal to the number of opcodes).
589
590 8. Following on from 7, I added a similar kludge to check the length of the
591 eint vector in pcreposix.c.
592
593 9. Error texts for pcre_compile() are held as one long string to avoid too
594 much relocation at load time. To find a text, the string is searched,
595 counting zeros. There was no check for running off the end of the string,
596 which could happen if a new error number was added without updating the
597 string.
598
599 10. \K gave a compile-time error if it appeared in a lookbehind assersion.
600
601 11. \K was not working if it appeared in an atomic group or in a group that
602 was called as a "subroutine", or in an assertion. Perl 5.11 documents that
603 \K is "not well defined" if used in an assertion. PCRE now accepts it if
604 the assertion is positive, but not if it is negative.
605
606 12. Change 11 fortuitously reduced the size of the stack frame used in the
607 "match()" function of pcre_exec.c by one pointer. Forthcoming
608 implementation of support for (*MARK) will need an extra pointer on the
609 stack; I have reserved it now, so that the stack frame size does not
610 decrease.
611
612 13. A pattern such as (?P<L1>(?P<L2>0)|(?P>L2)(?P>L1)) in which the only other
613 item in branch that calls a recursion is a subroutine call - as in the
614 second branch in the above example - was incorrectly given the compile-
615 time error "recursive call could loop indefinitely" because pcre_compile()
616 was not correctly checking the subroutine for matching a non-empty string.
617
618 14. The checks for overrunning compiling workspace could trigger after an
619 overrun had occurred. This is a "should never occur" error, but it can be
620 triggered by pathological patterns such as hundreds of nested parentheses.
621 The checks now trigger 100 bytes before the end of the workspace.
622
623 15. Fix typo in configure.ac: "srtoq" should be "strtoq".
624
625
626 Version 8.01 19-Jan-2010
627 ------------------------
628
629 1. If a pattern contained a conditional subpattern with only one branch (in
630 particular, this includes all (*DEFINE) patterns), a call to pcre_study()
631 computed the wrong minimum data length (which is of course zero for such
632 subpatterns). This could cause incorrect "no match" results.
633
634 2. For patterns such as (?i)a(?-i)b|c where an option setting at the start of
635 the pattern is reset in the first branch, pcre_compile() failed with
636 "internal error: code overflow at offset...". This happened only when
637 the reset was to the original external option setting. (An optimization
638 abstracts leading options settings into an external setting, which was the
639 cause of this.)
640
641 3. A pattern such as ^(?!a(*SKIP)b) where a negative assertion contained one
642 of the verbs SKIP, PRUNE, or COMMIT, did not work correctly. When the
643 assertion pattern did not match (meaning that the assertion was true), it
644 was incorrectly treated as false if the SKIP had been reached during the
645 matching. This also applied to assertions used as conditions.
646
647 4. If an item that is not supported by pcre_dfa_exec() was encountered in an
648 assertion subpattern, including such a pattern used as a condition,
649 unpredictable results occurred, instead of the error return
650 PCRE_ERROR_DFA_UITEM.
651
652 5. The C++ GlobalReplace function was not working like Perl for the special
653 situation when an empty string is matched. It now does the fancy magic
654 stuff that is necessary.
655
656 6. In pcre_internal.h, obsolete includes to setjmp.h and stdarg.h have been
657 removed. (These were left over from very, very early versions of PCRE.)
658
659 7. Some cosmetic changes to the code to make life easier when compiling it
660 as part of something else:
661
662 (a) Change DEBUG to PCRE_DEBUG.
663
664 (b) In pcre_compile(), rename the member of the "branch_chain" structure
665 called "current" as "current_branch", to prevent a collision with the
666 Linux macro when compiled as a kernel module.
667
668 (c) In pcre_study(), rename the function set_bit() as set_table_bit(), to
669 prevent a collision with the Linux macro when compiled as a kernel
670 module.
671
672 8. In pcre_compile() there are some checks for integer overflows that used to
673 cast potentially large values to (double). This has been changed to that
674 when building, a check for int64_t is made, and if it is found, it is used
675 instead, thus avoiding the use of floating point arithmetic. (There is no
676 other use of FP in PCRE.) If int64_t is not found, the fallback is to
677 double.
678
679 9. Added two casts to avoid signed/unsigned warnings from VS Studio Express
680 2005 (difference between two addresses compared to an unsigned value).
681
682 10. Change the standard AC_CHECK_LIB test for libbz2 in configure.ac to a
683 custom one, because of the following reported problem in Windows:
684
685 - libbz2 uses the Pascal calling convention (WINAPI) for the functions
686 under Win32.
687 - The standard autoconf AC_CHECK_LIB fails to include "bzlib.h",
688 therefore missing the function definition.
689 - The compiler thus generates a "C" signature for the test function.
690 - The linker fails to find the "C" function.
691 - PCRE fails to configure if asked to do so against libbz2.
692
693 11. When running libtoolize from libtool-2.2.6b as part of autogen.sh, these
694 messages were output:
695
696 Consider adding `AC_CONFIG_MACRO_DIR([m4])' to configure.ac and
697 rerunning libtoolize, to keep the correct libtool macros in-tree.
698 Consider adding `-I m4' to ACLOCAL_AMFLAGS in Makefile.am.
699
700 I have done both of these things.
701
702 12. Although pcre_dfa_exec() does not use nearly as much stack as pcre_exec()
703 most of the time, it *can* run out if it is given a pattern that contains a
704 runaway infinite recursion. I updated the discussion in the pcrestack man
705 page.
706
707 13. Now that we have gone to the x.xx style of version numbers, the minor
708 version may start with zero. Using 08 or 09 is a bad idea because users
709 might check the value of PCRE_MINOR in their code, and 08 or 09 may be
710 interpreted as invalid octal numbers. I've updated the previous comment in
711 configure.ac, and also added a check that gives an error if 08 or 09 are
712 used.
713
714 14. Change 8.00/11 was not quite complete: code had been accidentally omitted,
715 causing partial matching to fail when the end of the subject matched \W
716 in a UTF-8 pattern where \W was quantified with a minimum of 3.
717
718 15. There were some discrepancies between the declarations in pcre_internal.h
719 of _pcre_is_newline(), _pcre_was_newline(), and _pcre_valid_utf8() and
720 their definitions. The declarations used "const uschar *" and the
721 definitions used USPTR. Even though USPTR is normally defined as "const
722 unsigned char *" (and uschar is typedeffed as "unsigned char"), it was
723 reported that: "This difference in casting confuses some C++ compilers, for
724 example, SunCC recognizes above declarations as different functions and
725 generates broken code for hbpcre." I have changed the declarations to use
726 USPTR.
727
728 16. GNU libtool is named differently on some systems. The autogen.sh script now
729 tries several variants such as glibtoolize (MacOSX) and libtoolize1x
730 (FreeBSD).
731
732 17. Applied Craig's patch that fixes an HP aCC compile error in pcre 8.00
733 (strtoXX undefined when compiling pcrecpp.cc). The patch contains this
734 comment: "Figure out how to create a longlong from a string: strtoll and
735 equivalent. It's not enough to call AC_CHECK_FUNCS: hpux has a strtoll, for
736 instance, but it only takes 2 args instead of 3!"
737
738 18. A subtle bug concerned with back references has been fixed by a change of
739 specification, with a corresponding code fix. A pattern such as
740 ^(xa|=?\1a)+$ which contains a back reference inside the group to which it
741 refers, was giving matches when it shouldn't. For example, xa=xaaa would
742 match that pattern. Interestingly, Perl (at least up to 5.11.3) has the
743 same bug. Such groups have to be quantified to be useful, or contained
744 inside another quantified group. (If there's no repetition, the reference
745 can never match.) The problem arises because, having left the group and
746 moved on to the rest of the pattern, a later failure that backtracks into
747 the group uses the captured value from the final iteration of the group
748 rather than the correct earlier one. I have fixed this in PCRE by forcing
749 any group that contains a reference to itself to be an atomic group; that
750 is, there cannot be any backtracking into it once it has completed. This is
751 similar to recursive and subroutine calls.
752
753
754 Version 8.00 19-Oct-09
755 ----------------------
756
757 1. The table for translating pcre_compile() error codes into POSIX error codes
758 was out-of-date, and there was no check on the pcre_compile() error code
759 being within the table. This could lead to an OK return being given in
760 error.
761
762 2. Changed the call to open a subject file in pcregrep from fopen(pathname,
763 "r") to fopen(pathname, "rb"), which fixed a problem with some of the tests
764 in a Windows environment.
765
766 3. The pcregrep --count option prints the count for each file even when it is
767 zero, as does GNU grep. However, pcregrep was also printing all files when
768 --files-with-matches was added. Now, when both options are given, it prints
769 counts only for those files that have at least one match. (GNU grep just
770 prints the file name in this circumstance, but including the count seems
771 more useful - otherwise, why use --count?) Also ensured that the
772 combination -clh just lists non-zero counts, with no names.
773
774 4. The long form of the pcregrep -F option was incorrectly implemented as
775 --fixed_strings instead of --fixed-strings. This is an incompatible change,
776 but it seems right to fix it, and I didn't think it was worth preserving
777 the old behaviour.
778
779 5. The command line items --regex=pattern and --regexp=pattern were not
780 recognized by pcregrep, which required --regex pattern or --regexp pattern
781 (with a space rather than an '='). The man page documented the '=' forms,
782 which are compatible with GNU grep; these now work.
783
784 6. No libpcreposix.pc file was created for pkg-config; there was just
785 libpcre.pc and libpcrecpp.pc. The omission has been rectified.
786
787 7. Added #ifndef SUPPORT_UCP into the pcre_ucd.c module, to reduce its size
788 when UCP support is not needed, by modifying the Python script that
789 generates it from Unicode data files. This should not matter if the module
790 is correctly used as a library, but I received one complaint about 50K of
791 unwanted data. My guess is that the person linked everything into his
792 program rather than using a library. Anyway, it does no harm.
793
794 8. A pattern such as /\x{123}{2,2}+/8 was incorrectly compiled; the trigger
795 was a minimum greater than 1 for a wide character in a possessive
796 repetition. The same bug could also affect patterns like /(\x{ff}{0,2})*/8
797 which had an unlimited repeat of a nested, fixed maximum repeat of a wide
798 character. Chaos in the form of incorrect output or a compiling loop could
799 result.
800
801 9. The restrictions on what a pattern can contain when partial matching is
802 requested for pcre_exec() have been removed. All patterns can now be
803 partially matched by this function. In addition, if there are at least two
804 slots in the offset vector, the offset of the earliest inspected character
805 for the match and the offset of the end of the subject are set in them when
806 PCRE_ERROR_PARTIAL is returned.
807
808 10. Partial matching has been split into two forms: PCRE_PARTIAL_SOFT, which is
809 synonymous with PCRE_PARTIAL, for backwards compatibility, and
810 PCRE_PARTIAL_HARD, which causes a partial match to supersede a full match,
811 and may be more useful for multi-segment matching.
812
813 11. Partial matching with pcre_exec() is now more intuitive. A partial match
814 used to be given if ever the end of the subject was reached; now it is
815 given only if matching could not proceed because another character was
816 needed. This makes a difference in some odd cases such as Z(*FAIL) with the
817 string "Z", which now yields "no match" instead of "partial match". In the
818 case of pcre_dfa_exec(), "no match" is given if every matching path for the
819 final character ended with (*FAIL).
820
821 12. Restarting a match using pcre_dfa_exec() after a partial match did not work
822 if the pattern had a "must contain" character that was already found in the
823 earlier partial match, unless partial matching was again requested. For
824 example, with the pattern /dog.(body)?/, the "must contain" character is
825 "g". If the first part-match was for the string "dog", restarting with
826 "sbody" failed. This bug has been fixed.
827
828 13. The string returned by pcre_dfa_exec() after a partial match has been
829 changed so that it starts at the first inspected character rather than the
830 first character of the match. This makes a difference only if the pattern
831 starts with a lookbehind assertion or \b or \B (\K is not supported by
832 pcre_dfa_exec()). It's an incompatible change, but it makes the two
833 matching functions compatible, and I think it's the right thing to do.
834
835 14. Added a pcredemo man page, created automatically from the pcredemo.c file,
836 so that the demonstration program is easily available in environments where
837 PCRE has not been installed from source.
838
839 15. Arranged to add -DPCRE_STATIC to cflags in libpcre.pc, libpcreposix.cp,
840 libpcrecpp.pc and pcre-config when PCRE is not compiled as a shared
841 library.
842
843 16. Added REG_UNGREEDY to the pcreposix interface, at the request of a user.
844 It maps to PCRE_UNGREEDY. It is not, of course, POSIX-compatible, but it
845 is not the first non-POSIX option to be added. Clearly some people find
846 these options useful.
847
848 17. If a caller to the POSIX matching function regexec() passes a non-zero
849 value for nmatch with a NULL value for pmatch, the value of
850 nmatch is forced to zero.
851
852 18. RunGrepTest did not have a test for the availability of the -u option of
853 the diff command, as RunTest does. It now checks in the same way as
854 RunTest, and also checks for the -b option.
855
856 19. If an odd number of negated classes containing just a single character
857 interposed, within parentheses, between a forward reference to a named
858 subpattern and the definition of the subpattern, compilation crashed with
859 an internal error, complaining that it could not find the referenced
860 subpattern. An example of a crashing pattern is /(?&A)(([^m])(?<A>))/.
861 [The bug was that it was starting one character too far in when skipping
862 over the character class, thus treating the ] as data rather than
863 terminating the class. This meant it could skip too much.]
864
865 20. Added PCRE_NOTEMPTY_ATSTART in order to be able to correctly implement the
866 /g option in pcretest when the pattern contains \K, which makes it possible
867 to have an empty string match not at the start, even when the pattern is
868 anchored. Updated pcretest and pcredemo to use this option.
869
870 21. If the maximum number of capturing subpatterns in a recursion was greater
871 than the maximum at the outer level, the higher number was returned, but
872 with unset values at the outer level. The correct (outer level) value is
873 now given.
874
875 22. If (*ACCEPT) appeared inside capturing parentheses, previous releases of
876 PCRE did not set those parentheses (unlike Perl). I have now found a way to
877 make it do so. The string so far is captured, making this feature
878 compatible with Perl.
879
880 23. The tests have been re-organized, adding tests 11 and 12, to make it
881 possible to check the Perl 5.10 features against Perl 5.10.
882
883 24. Perl 5.10 allows subroutine calls in lookbehinds, as long as the subroutine
884 pattern matches a fixed length string. PCRE did not allow this; now it
885 does. Neither allows recursion.
886
887 25. I finally figured out how to implement a request to provide the minimum
888 length of subject string that was needed in order to match a given pattern.
889 (It was back references and recursion that I had previously got hung up
890 on.) This code has now been added to pcre_study(); it finds a lower bound
891 to the length of subject needed. It is not necessarily the greatest lower
892 bound, but using it to avoid searching strings that are too short does give
893 some useful speed-ups. The value is available to calling programs via
894 pcre_fullinfo().
895
896 26. While implementing 25, I discovered to my embarrassment that pcretest had
897 not been passing the result of pcre_study() to pcre_dfa_exec(), so the
898 study optimizations had never been tested with that matching function.
899 Oops. What is worse, even when it was passed study data, there was a bug in
900 pcre_dfa_exec() that meant it never actually used it. Double oops. There
901 were also very few tests of studied patterns with pcre_dfa_exec().
902
903 27. If (?| is used to create subpatterns with duplicate numbers, they are now
904 allowed to have the same name, even if PCRE_DUPNAMES is not set. However,
905 on the other side of the coin, they are no longer allowed to have different
906 names, because these cannot be distinguished in PCRE, and this has caused
907 confusion. (This is a difference from Perl.)
908
909 28. When duplicate subpattern names are present (necessarily with different
910 numbers, as required by 27 above), and a test is made by name in a
911 conditional pattern, either for a subpattern having been matched, or for
912 recursion in such a pattern, all the associated numbered subpatterns are
913 tested, and the overall condition is true if the condition is true for any
914 one of them. This is the way Perl works, and is also more like the way
915 testing by number works.
916
917
918 Version 7.9 11-Apr-09
919 ---------------------
920
921 1. When building with support for bzlib/zlib (pcregrep) and/or readline
922 (pcretest), all targets were linked against these libraries. This included
923 libpcre, libpcreposix, and libpcrecpp, even though they do not use these
924 libraries. This caused unwanted dependencies to be created. This problem
925 has been fixed, and now only pcregrep is linked with bzlib/zlib and only
926 pcretest is linked with readline.
927
928 2. The "typedef int BOOL" in pcre_internal.h that was included inside the
929 "#ifndef FALSE" condition by an earlier change (probably 7.8/18) has been
930 moved outside it again, because FALSE and TRUE are already defined in AIX,
931 but BOOL is not.
932
933 3. The pcre_config() function was treating the PCRE_MATCH_LIMIT and
934 PCRE_MATCH_LIMIT_RECURSION values as ints, when they should be long ints.
935
936 4. The pcregrep documentation said spaces were inserted as well as colons (or
937 hyphens) following file names and line numbers when outputting matching
938 lines. This is not true; no spaces are inserted. I have also clarified the
939 wording for the --colour (or --color) option.
940
941 5. In pcregrep, when --colour was used with -o, the list of matching strings
942 was not coloured; this is different to GNU grep, so I have changed it to be
943 the same.
944
945 6. When --colo(u)r was used in pcregrep, only the first matching substring in
946 each matching line was coloured. Now it goes on to look for further matches
947 of any of the test patterns, which is the same behaviour as GNU grep.
948
949 7. A pattern that could match an empty string could cause pcregrep to loop; it
950 doesn't make sense to accept an empty string match in pcregrep, so I have
951 locked it out (using PCRE's PCRE_NOTEMPTY option). By experiment, this
952 seems to be how GNU grep behaves.
953
954 8. The pattern (?(?=.*b)b|^) was incorrectly compiled as "match must be at
955 start or after a newline", because the conditional assertion was not being
956 correctly handled. The rule now is that both the assertion and what follows
957 in the first alternative must satisfy the test.
958
959 9. If auto-callout was enabled in a pattern with a conditional group whose
960 condition was an assertion, PCRE could crash during matching, both with
961 pcre_exec() and pcre_dfa_exec().
962
963 10. The PCRE_DOLLAR_ENDONLY option was not working when pcre_dfa_exec() was
964 used for matching.
965
966 11. Unicode property support in character classes was not working for
967 characters (bytes) greater than 127 when not in UTF-8 mode.
968
969 12. Added the -M command line option to pcretest.
970
971 14. Added the non-standard REG_NOTEMPTY option to the POSIX interface.
972
973 15. Added the PCRE_NO_START_OPTIMIZE match-time option.
974
975 16. Added comments and documentation about mis-use of no_arg in the C++
976 wrapper.
977
978 17. Implemented support for UTF-8 encoding in EBCDIC environments, a patch
979 from Martin Jerabek that uses macro names for all relevant character and
980 string constants.
981
982 18. Added to pcre_internal.h two configuration checks: (a) If both EBCDIC and
983 SUPPORT_UTF8 are set, give an error; (b) If SUPPORT_UCP is set without
984 SUPPORT_UTF8, define SUPPORT_UTF8. The "configure" script handles both of
985 these, but not everybody uses configure.
986
987 19. A conditional group that had only one branch was not being correctly
988 recognized as an item that could match an empty string. This meant that an
989 enclosing group might also not be so recognized, causing infinite looping
990 (and probably a segfault) for patterns such as ^"((?(?=[a])[^"])|b)*"$
991 with the subject "ab", where knowledge that the repeated group can match
992 nothing is needed in order to break the loop.
993
994 20. If a pattern that was compiled with callouts was matched using pcre_dfa_
995 exec(), but without supplying a callout function, matching went wrong.
996
997 21. If PCRE_ERROR_MATCHLIMIT occurred during a recursion, there was a memory
998 leak if the size of the offset vector was greater than 30. When the vector
999 is smaller, the saved offsets during recursion go onto a local stack
1000 vector, but for larger vectors malloc() is used. It was failing to free
1001 when the recursion yielded PCRE_ERROR_MATCH_LIMIT (or any other "abnormal"
1002 error, in fact).
1003
1004 22. There was a missing #ifdef SUPPORT_UTF8 round one of the variables in the
1005 heapframe that is used only when UTF-8 support is enabled. This caused no
1006 problem, but was untidy.
1007
1008 23. Steven Van Ingelgem's patch to CMakeLists.txt to change the name
1009 CMAKE_BINARY_DIR to PROJECT_BINARY_DIR so that it works when PCRE is
1010 included within another project.
1011
1012 24. Steven Van Ingelgem's patches to add more options to the CMake support,
1013 slightly modified by me:
1014
1015 (a) PCRE_BUILD_TESTS can be set OFF not to build the tests, including
1016 not building pcregrep.
1017
1018 (b) PCRE_BUILD_PCREGREP can be see OFF not to build pcregrep, but only
1019 if PCRE_BUILD_TESTS is also set OFF, because the tests use pcregrep.
1020
1021 25. Forward references, both numeric and by name, in patterns that made use of
1022 duplicate group numbers, could behave incorrectly or give incorrect errors,
1023 because when scanning forward to find the reference group, PCRE was not
1024 taking into account the duplicate group numbers. A pattern such as
1025 ^X(?3)(a)(?|(b)|(q))(Y) is an example.
1026
1027 26. Changed a few more instances of "const unsigned char *" to USPTR, making
1028 the feature of a custom pointer more persuasive (as requested by a user).
1029
1030 27. Wrapped the definitions of fileno and isatty for Windows, which appear in
1031 pcretest.c, inside #ifndefs, because it seems they are sometimes already
1032 pre-defined.
1033
1034 28. Added support for (*UTF8) at the start of a pattern.
1035
1036 29. Arrange for flags added by the "release type" setting in CMake to be shown
1037 in the configuration summary.
1038
1039
1040 Version 7.8 05-Sep-08
1041 ---------------------
1042
1043 1. Replaced UCP searching code with optimized version as implemented for Ad
1044 Muncher (http://www.admuncher.com/) by Peter Kankowski. This uses a two-
1045 stage table and inline lookup instead of a function, giving speed ups of 2
1046 to 5 times on some simple patterns that I tested. Permission was given to
1047 distribute the MultiStage2.py script that generates the tables (it's not in
1048 the tarball, but is in the Subversion repository).
1049
1050 2. Updated the Unicode datatables to Unicode 5.1.0. This adds yet more
1051 scripts.
1052
1053 3. Change 12 for 7.7 introduced a bug in pcre_study() when a pattern contained
1054 a group with a zero qualifier. The result of the study could be incorrect,
1055 or the function might crash, depending on the pattern.
1056
1057 4. Caseless matching was not working for non-ASCII characters in back
1058 references. For example, /(\x{de})\1/8i was not matching \x{de}\x{fe}.
1059 It now works when Unicode Property Support is available.
1060
1061 5. In pcretest, an escape such as \x{de} in the data was always generating
1062 a UTF-8 string, even in non-UTF-8 mode. Now it generates a single byte in
1063 non-UTF-8 mode. If the value is greater than 255, it gives a warning about
1064 truncation.
1065
1066 6. Minor bugfix in pcrecpp.cc (change "" == ... to NULL == ...).
1067
1068 7. Added two (int) casts to pcregrep when printing the difference of two
1069 pointers, in case they are 64-bit values.
1070
1071 8. Added comments about Mac OS X stack usage to the pcrestack man page and to
1072 test 2 if it fails.
1073
1074 9. Added PCRE_CALL_CONVENTION just before the names of all exported functions,
1075 and a #define of that name to empty if it is not externally set. This is to
1076 allow users of MSVC to set it if necessary.
1077
1078 10. The PCRE_EXP_DEFN macro which precedes exported functions was missing from
1079 the convenience functions in the pcre_get.c source file.
1080
1081 11. An option change at the start of a pattern that had top-level alternatives
1082 could cause overwriting and/or a crash. This command provoked a crash in
1083 some environments:
1084
1085 printf "/(?i)[\xc3\xa9\xc3\xbd]|[\xc3\xa9\xc3\xbdA]/8\n" | pcretest
1086
1087 This potential security problem was recorded as CVE-2008-2371.
1088
1089 12. For a pattern where the match had to start at the beginning or immediately
1090 after a newline (e.g /.*anything/ without the DOTALL flag), pcre_exec() and
1091 pcre_dfa_exec() could read past the end of the passed subject if there was
1092 no match. To help with detecting such bugs (e.g. with valgrind), I modified
1093 pcretest so that it places the subject at the end of its malloc-ed buffer.
1094
1095 13. The change to pcretest in 12 above threw up a couple more cases when pcre_
1096 exec() might read past the end of the data buffer in UTF-8 mode.
1097
1098 14. A similar bug to 7.3/2 existed when the PCRE_FIRSTLINE option was set and
1099 the data contained the byte 0x85 as part of a UTF-8 character within its
1100 first line. This applied both to normal and DFA matching.
1101
1102 15. Lazy qualifiers were not working in some cases in UTF-8 mode. For example,
1103 /^[^d]*?$/8 failed to match "abc".
1104
1105 16. Added a missing copyright notice to pcrecpp_internal.h.
1106
1107 17. Make it more clear in the documentation that values returned from
1108 pcre_exec() in ovector are byte offsets, not character counts.
1109
1110 18. Tidied a few places to stop certain compilers from issuing warnings.
1111
1112 19. Updated the Virtual Pascal + BCC files to compile the latest v7.7, as
1113 supplied by Stefan Weber. I made a further small update for 7.8 because
1114 there is a change of source arrangements: the pcre_searchfuncs.c module is
1115 replaced by pcre_ucd.c.
1116
1117
1118 Version 7.7 07-May-08
1119 ---------------------
1120
1121 1. Applied Craig's patch to sort out a long long problem: "If we can't convert
1122 a string to a long long, pretend we don't even have a long long." This is
1123 done by checking for the strtoq, strtoll, and _strtoi64 functions.
1124
1125 2. Applied Craig's patch to pcrecpp.cc to restore ABI compatibility with
1126 pre-7.6 versions, which defined a global no_arg variable instead of putting
1127 it in the RE class. (See also #8 below.)
1128
1129 3. Remove a line of dead code, identified by coverity and reported by Nuno
1130 Lopes.
1131
1132 4. Fixed two related pcregrep bugs involving -r with --include or --exclude:
1133
1134 (1) The include/exclude patterns were being applied to the whole pathnames
1135 of files, instead of just to the final components.
1136
1137 (2) If there was more than one level of directory, the subdirectories were
1138 skipped unless they satisfied the include/exclude conditions. This is
1139 inconsistent with GNU grep (and could even be seen as contrary to the
1140 pcregrep specification - which I improved to make it absolutely clear).
1141 The action now is always to scan all levels of directory, and just
1142 apply the include/exclude patterns to regular files.
1143
1144 5. Added the --include_dir and --exclude_dir patterns to pcregrep, and used
1145 --exclude_dir in the tests to avoid scanning .svn directories.
1146
1147 6. Applied Craig's patch to the QuoteMeta function so that it escapes the
1148 NUL character as backslash + 0 rather than backslash + NUL, because PCRE
1149 doesn't support NULs in patterns.
1150
1151 7. Added some missing "const"s to declarations of static tables in
1152 pcre_compile.c and pcre_dfa_exec.c.
1153
1154 8. Applied Craig's patch to pcrecpp.cc to fix a problem in OS X that was
1155 caused by fix #2 above. (Subsequently also a second patch to fix the
1156 first patch. And a third patch - this was a messy problem.)
1157
1158 9. Applied Craig's patch to remove the use of push_back().
1159
1160 10. Applied Alan Lehotsky's patch to add REG_STARTEND support to the POSIX
1161 matching function regexec().
1162
1163 11. Added support for the Oniguruma syntax \g<name>, \g<n>, \g'name', \g'n',
1164 which, however, unlike Perl's \g{...}, are subroutine calls, not back
1165 references. PCRE supports relative numbers with this syntax (I don't think
1166 Oniguruma does).
1167
1168 12. Previously, a group with a zero repeat such as (...){0} was completely
1169 omitted from the compiled regex. However, this means that if the group
1170 was called as a subroutine from elsewhere in the pattern, things went wrong
1171 (an internal error was given). Such groups are now left in the compiled
1172 pattern, with a new opcode that causes them to be skipped at execution
1173 time.
1174
1175 13. Added the PCRE_JAVASCRIPT_COMPAT option. This makes the following changes
1176 to the way PCRE behaves:
1177
1178 (a) A lone ] character is dis-allowed (Perl treats it as data).
1179
1180 (b) A back reference to an unmatched subpattern matches an empty string
1181 (Perl fails the current match path).
1182
1183 (c) A data ] in a character class must be notated as \] because if the
1184 first data character in a class is ], it defines an empty class. (In
1185 Perl it is not possible to have an empty class.) The empty class []
1186 never matches; it forces failure and is equivalent to (*FAIL) or (?!).
1187 The negative empty class [^] matches any one character, independently
1188 of the DOTALL setting.
1189
1190 14. A pattern such as /(?2)[]a()b](abc)/ which had a forward reference to a
1191 non-existent subpattern following a character class starting with ']' and
1192 containing () gave an internal compiling error instead of "reference to
1193 non-existent subpattern". Fortunately, when the pattern did exist, the
1194 compiled code was correct. (When scanning forwards to check for the
1195 existencd of the subpattern, it was treating the data ']' as terminating
1196 the class, so got the count wrong. When actually compiling, the reference
1197 was subsequently set up correctly.)
1198
1199 15. The "always fail" assertion (?!) is optimzed to (*FAIL) by pcre_compile;
1200 it was being rejected as not supported by pcre_dfa_exec(), even though
1201 other assertions are supported. I have made pcre_dfa_exec() support
1202 (*FAIL).
1203
1204 16. The implementation of 13c above involved the invention of a new opcode,
1205 OP_ALLANY, which is like OP_ANY but doesn't check the /s flag. Since /s
1206 cannot be changed at match time, I realized I could make a small
1207 improvement to matching performance by compiling OP_ALLANY instead of
1208 OP_ANY for "." when DOTALL was set, and then removing the runtime tests
1209 on the OP_ANY path.
1210
1211 17. Compiling pcretest on Windows with readline support failed without the
1212 following two fixes: (1) Make the unistd.h include conditional on
1213 HAVE_UNISTD_H; (2) #define isatty and fileno as _isatty and _fileno.
1214
1215 18. Changed CMakeLists.txt and cmake/FindReadline.cmake to arrange for the
1216 ncurses library to be included for pcretest when ReadLine support is
1217 requested, but also to allow for it to be overridden. This patch came from
1218 Daniel Bergström.
1219
1220 19. There was a typo in the file ucpinternal.h where f0_rangeflag was defined
1221 as 0x00f00000 instead of 0x00800000. Luckily, this would not have caused
1222 any errors with the current Unicode tables. Thanks to Peter Kankowski for
1223 spotting this.
1224
1225
1226 Version 7.6 28-Jan-08
1227 ---------------------
1228
1229 1. A character class containing a very large number of characters with
1230 codepoints greater than 255 (in UTF-8 mode, of course) caused a buffer
1231 overflow.
1232
1233 2. Patch to cut out the "long long" test in pcrecpp_unittest when
1234 HAVE_LONG_LONG is not defined.
1235
1236 3. Applied Christian Ehrlicher's patch to update the CMake build files to
1237 bring them up to date and include new features. This patch includes:
1238
1239 - Fixed PH's badly added libz and libbz2 support.
1240 - Fixed a problem with static linking.
1241 - Added pcredemo. [But later removed - see 7 below.]
1242 - Fixed dftables problem and added an option.
1243 - Added a number of HAVE_XXX tests, including HAVE_WINDOWS_H and
1244 HAVE_LONG_LONG.
1245 - Added readline support for pcretest.
1246 - Added an listing of the option settings after cmake has run.
1247
1248 4. A user submitted a patch to Makefile that makes it easy to create
1249 "pcre.dll" under mingw when using Configure/Make. I added stuff to
1250 Makefile.am that cause it to include this special target, without
1251 affecting anything else. Note that the same mingw target plus all
1252 the other distribution libraries and programs are now supported
1253 when configuring with CMake (see 6 below) instead of with
1254 Configure/Make.
1255
1256 5. Applied Craig's patch that moves no_arg into the RE class in the C++ code.
1257 This is an attempt to solve the reported problem "pcrecpp::no_arg is not
1258 exported in the Windows port". It has not yet been confirmed that the patch
1259 solves the problem, but it does no harm.
1260
1261 6. Applied Sheri's patch to CMakeLists.txt to add NON_STANDARD_LIB_PREFIX and
1262 NON_STANDARD_LIB_SUFFIX for dll names built with mingw when configured
1263 with CMake, and also correct the comment about stack recursion.
1264
1265 7. Remove the automatic building of pcredemo from the ./configure system and
1266 from CMakeLists.txt. The whole idea of pcredemo.c is that it is an example
1267 of a program that users should build themselves after PCRE is installed, so
1268 building it automatically is not really right. What is more, it gave
1269 trouble in some build environments.
1270
1271 8. Further tidies to CMakeLists.txt from Sheri and Christian.
1272
1273
1274 Version 7.5 10-Jan-08
1275 ---------------------
1276
1277 1. Applied a patch from Craig: "This patch makes it possible to 'ignore'
1278 values in parens when parsing an RE using the C++ wrapper."
1279
1280 2. Negative specials like \S did not work in character classes in UTF-8 mode.
1281 Characters greater than 255 were excluded from the class instead of being
1282 included.
1283
1284 3. The same bug as (2) above applied to negated POSIX classes such as
1285 [:^space:].
1286
1287 4. PCRECPP_STATIC was referenced in pcrecpp_internal.h, but nowhere was it
1288 defined or documented. It seems to have been a typo for PCRE_STATIC, so
1289 I have changed it.
1290
1291 5. The construct (?&) was not diagnosed as a syntax error (it referenced the
1292 first named subpattern) and a construct such as (?&a) would reference the
1293 first named subpattern whose name started with "a" (in other words, the
1294 length check was missing). Both these problems are fixed. "Subpattern name
1295 expected" is now given for (?&) (a zero-length name), and this patch also
1296 makes it give the same error for \k'' (previously it complained that that
1297 was a reference to a non-existent subpattern).
1298
1299 6. The erroneous patterns (?+-a) and (?-+a) give different error messages;
1300 this is right because (?- can be followed by option settings as well as by
1301 digits. I have, however, made the messages clearer.
1302
1303 7. Patterns such as (?(1)a|b) (a pattern that contains fewer subpatterns
1304 than the number used in the conditional) now cause a compile-time error.
1305 This is actually not compatible with Perl, which accepts such patterns, but
1306 treats the conditional as always being FALSE (as PCRE used to), but it
1307 seems to me that giving a diagnostic is better.
1308
1309 8. Change "alphameric" to the more common word "alphanumeric" in comments
1310 and messages.
1311
1312 9. Fix two occurrences of "backslash" in comments that should have been
1313 "backspace".
1314
1315 10. Remove two redundant lines of code that can never be obeyed (their function
1316 was moved elsewhere).
1317
1318 11. The program that makes PCRE's Unicode character property table had a bug
1319 which caused it to generate incorrect table entries for sequences of
1320 characters that have the same character type, but are in different scripts.
1321 It amalgamated them into a single range, with the script of the first of
1322 them. In other words, some characters were in the wrong script. There were
1323 thirteen such cases, affecting characters in the following ranges:
1324
1325 U+002b0 - U+002c1
1326 U+0060c - U+0060d
1327 U+0061e - U+00612
1328 U+0064b - U+0065e
1329 U+0074d - U+0076d
1330 U+01800 - U+01805
1331 U+01d00 - U+01d77
1332 U+01d9b - U+01dbf
1333 U+0200b - U+0200f
1334 U+030fc - U+030fe
1335 U+03260 - U+0327f
1336 U+0fb46 - U+0fbb1
1337 U+10450 - U+1049d
1338
1339 12. The -o option (show only the matching part of a line) for pcregrep was not
1340 compatible with GNU grep in that, if there was more than one match in a
1341 line, it showed only the first of them. It now behaves in the same way as
1342 GNU grep.
1343
1344 13. If the -o and -v options were combined for pcregrep, it printed a blank
1345 line for every non-matching line. GNU grep prints nothing, and pcregrep now
1346 does the same. The return code can be used to tell if there were any
1347 non-matching lines.
1348
1349 14. Added --file-offsets and --line-offsets to pcregrep.
1350
1351 15. The pattern (?=something)(?R) was not being diagnosed as a potentially
1352 infinitely looping recursion. The bug was that positive lookaheads were not
1353 being skipped when checking for a possible empty match (negative lookaheads
1354 and both kinds of lookbehind were skipped).
1355
1356 16. Fixed two typos in the Windows-only code in pcregrep.c, and moved the
1357 inclusion of <windows.h> to before rather than after the definition of
1358 INVALID_FILE_ATTRIBUTES (patch from David Byron).
1359
1360 17. Specifying a possessive quantifier with a specific limit for a Unicode
1361 character property caused pcre_compile() to compile bad code, which led at
1362 runtime to PCRE_ERROR_INTERNAL (-14). Examples of patterns that caused this
1363 are: /\p{Zl}{2,3}+/8 and /\p{Cc}{2}+/8. It was the possessive "+" that
1364 caused the error; without that there was no problem.
1365
1366 18. Added --enable-pcregrep-libz and --enable-pcregrep-libbz2.
1367
1368 19. Added --enable-pcretest-libreadline.
1369
1370 20. In pcrecpp.cc, the variable 'count' was incremented twice in
1371 RE::GlobalReplace(). As a result, the number of replacements returned was
1372 double what it should be. I removed one of the increments, but Craig sent a
1373 later patch that removed the other one (the right fix) and added unit tests
1374 that check the return values (which was not done before).
1375
1376 21. Several CMake things:
1377
1378 (1) Arranged that, when cmake is used on Unix, the libraries end up with
1379 the names libpcre and libpcreposix, not just pcre and pcreposix.
1380
1381 (2) The above change means that pcretest and pcregrep are now correctly
1382 linked with the newly-built libraries, not previously installed ones.
1383
1384 (3) Added PCRE_SUPPORT_LIBREADLINE, PCRE_SUPPORT_LIBZ, PCRE_SUPPORT_LIBBZ2.
1385
1386 22. In UTF-8 mode, with newline set to "any", a pattern such as .*a.*=.b.*
1387 crashed when matching a string such as a\x{2029}b (note that \x{2029} is a
1388 UTF-8 newline character). The key issue is that the pattern starts .*;
1389 this means that the match must be either at the beginning, or after a
1390 newline. The bug was in the code for advancing after a failed match and
1391 checking that the new position followed a newline. It was not taking
1392 account of UTF-8 characters correctly.
1393
1394 23. PCRE was behaving differently from Perl in the way it recognized POSIX
1395 character classes. PCRE was not treating the sequence [:...:] as a
1396 character class unless the ... were all letters. Perl, however, seems to
1397 allow any characters between [: and :], though of course it rejects as
1398 unknown any "names" that contain non-letters, because all the known class
1399 names consist only of letters. Thus, Perl gives an error for [[:1234:]],
1400 for example, whereas PCRE did not - it did not recognize a POSIX character
1401 class. This seemed a bit dangerous, so the code has been changed to be
1402 closer to Perl. The behaviour is not identical to Perl, because PCRE will
1403 diagnose an unknown class for, for example, [[:l\ower:]] where Perl will
1404 treat it as [[:lower:]]. However, PCRE does now give "unknown" errors where
1405 Perl does, and where it didn't before.
1406
1407 24. Rewrite so as to remove the single use of %n from pcregrep because in some
1408 Windows environments %n is disabled by default.
1409
1410
1411 Version 7.4 21-Sep-07
1412 ---------------------
1413
1414 1. Change 7.3/28 was implemented for classes by looking at the bitmap. This
1415 means that a class such as [\s] counted as "explicit reference to CR or
1416 LF". That isn't really right - the whole point of the change was to try to
1417 help when there was an actual mention of one of the two characters. So now
1418 the change happens only if \r or \n (or a literal CR or LF) character is
1419 encountered.
1420
1421 2. The 32-bit options word was also used for 6 internal flags, but the numbers
1422 of both had grown to the point where there were only 3 bits left.
1423 Fortunately, there was spare space in the data structure, and so I have
1424 moved the internal flags into a new 16-bit field to free up more option
1425 bits.
1426
1427 3. The appearance of (?J) at the start of a pattern set the DUPNAMES option,
1428 but did not set the internal JCHANGED flag - either of these is enough to
1429 control the way the "get" function works - but the PCRE_INFO_JCHANGED
1430 facility is supposed to tell if (?J) was ever used, so now (?J) at the
1431 start sets both bits.
1432
1433 4. Added options (at build time, compile time, exec time) to change \R from
1434 matching any Unicode line ending sequence to just matching CR, LF, or CRLF.
1435
1436 5. doc/pcresyntax.html was missing from the distribution.
1437
1438 6. Put back the definition of PCRE_ERROR_NULLWSLIMIT, for backward
1439 compatibility, even though it is no longer used.
1440
1441 7. Added macro for snprintf to pcrecpp_unittest.cc and also for strtoll and
1442 strtoull to pcrecpp.cc to select the available functions in WIN32 when the
1443 windows.h file is present (where different names are used). [This was
1444 reversed later after testing - see 16 below.]
1445
1446 8. Changed all #include <config.h> to #include "config.h". There were also
1447 some further <pcre.h> cases that I changed to "pcre.h".
1448
1449 9. When pcregrep was used with the --colour option, it missed the line ending
1450 sequence off the lines that it output.
1451
1452 10. It was pointed out to me that arrays of string pointers cause lots of
1453 relocations when a shared library is dynamically loaded. A technique of
1454 using a single long string with a table of offsets can drastically reduce
1455 these. I have refactored PCRE in four places to do this. The result is
1456 dramatic:
1457
1458 Originally: 290
1459 After changing UCP table: 187
1460 After changing error message table: 43
1461 After changing table of "verbs" 36
1462 After changing table of Posix names 22
1463
1464 Thanks to the folks working on Gregex for glib for this insight.
1465
1466 11. --disable-stack-for-recursion caused compiling to fail unless -enable-
1467 unicode-properties was also set.
1468
1469 12. Updated the tests so that they work when \R is defaulted to ANYCRLF.
1470
1471 13. Added checks for ANY and ANYCRLF to pcrecpp.cc where it previously
1472 checked only for CRLF.
1473
1474 14. Added casts to pcretest.c to avoid compiler warnings.
1475
1476 15. Added Craig's patch to various pcrecpp modules to avoid compiler warnings.
1477
1478 16. Added Craig's patch to remove the WINDOWS_H tests, that were not working,
1479 and instead check for _strtoi64 explicitly, and avoid the use of snprintf()
1480 entirely. This removes changes made in 7 above.
1481
1482 17. The CMake files have been updated, and there is now more information about
1483 building with CMake in the NON-UNIX-USE document.
1484
1485
1486 Version 7.3 28-Aug-07
1487 ---------------------
1488
1489 1. In the rejigging of the build system that eventually resulted in 7.1, the
1490 line "#include <pcre.h>" was included in pcre_internal.h. The use of angle
1491 brackets there is not right, since it causes compilers to look for an
1492 installed pcre.h, not the version that is in the source that is being
1493 compiled (which of course may be different). I have changed it back to:
1494
1495 #include "pcre.h"
1496
1497 I have a vague recollection that the change was concerned with compiling in
1498 different directories, but in the new build system, that is taken care of
1499 by the VPATH setting the Makefile.
1500
1501 2. The pattern .*$ when run in not-DOTALL UTF-8 mode with newline=any failed
1502 when the subject happened to end in the byte 0x85 (e.g. if the last
1503 character was \x{1ec5}). *Character* 0x85 is one of the "any" newline
1504 characters but of course it shouldn't be taken as a newline when it is part
1505 of another character. The bug was that, for an unlimited repeat of . in
1506 not-DOTALL UTF-8 mode, PCRE was advancing by bytes rather than by
1507 characters when looking for a newline.
1508
1509 3. A small performance improvement in the DOTALL UTF-8 mode .* case.
1510
1511 4. Debugging: adjusted the names of opcodes for different kinds of parentheses
1512 in debug output.
1513
1514 5. Arrange to use "%I64d" instead of "%lld" and "%I64u" instead of "%llu" for
1515 long printing in the pcrecpp unittest when running under MinGW.
1516
1517 6. ESC_K was left out of the EBCDIC table.
1518
1519 7. Change 7.0/38 introduced a new limit on the number of nested non-capturing
1520 parentheses; I made it 1000, which seemed large enough. Unfortunately, the
1521 limit also applies to "virtual nesting" when a pattern is recursive, and in
1522 this case 1000 isn't so big. I have been able to remove this limit at the
1523 expense of backing off one optimization in certain circumstances. Normally,
1524 when pcre_exec() would call its internal match() function recursively and
1525 immediately return the result unconditionally, it uses a "tail recursion"
1526 feature to save stack. However, when a subpattern that can match an empty
1527 string has an unlimited repetition quantifier, it no longer makes this
1528 optimization. That gives it a stack frame in which to save the data for
1529 checking that an empty string has been matched. Previously this was taken
1530 from the 1000-entry workspace that had been reserved. So now there is no
1531 explicit limit, but more stack is used.
1532
1533 8. Applied Daniel's patches to solve problems with the import/export magic
1534 syntax that is required for Windows, and which was going wrong for the
1535 pcreposix and pcrecpp parts of the library. These were overlooked when this
1536 problem was solved for the main library.
1537
1538 9. There were some crude static tests to avoid integer overflow when computing
1539 the size of patterns that contain repeated groups with explicit upper
1540 limits. As the maximum quantifier is 65535, the maximum group length was
1541 set at 30,000 so that the product of these two numbers did not overflow a
1542 32-bit integer. However, it turns out that people want to use groups that
1543 are longer than 30,000 bytes (though not repeat them that many times).
1544 Change 7.0/17 (the refactoring of the way the pattern size is computed) has
1545 made it possible to implement the integer overflow checks in a much more
1546 dynamic way, which I have now done. The artificial limitation on group
1547 length has been removed - we now have only the limit on the total length of
1548 the compiled pattern, which depends on the LINK_SIZE setting.
1549
1550 10. Fixed a bug in the documentation for get/copy named substring when
1551 duplicate names are permitted. If none of the named substrings are set, the
1552 functions return PCRE_ERROR_NOSUBSTRING (7); the doc said they returned an
1553 empty string.
1554
1555 11. Because Perl interprets \Q...\E at a high level, and ignores orphan \E
1556 instances, patterns such as [\Q\E] or [\E] or even [^\E] cause an error,
1557 because the ] is interpreted as the first data character and the
1558 terminating ] is not found. PCRE has been made compatible with Perl in this
1559 regard. Previously, it interpreted [\Q\E] as an empty class, and [\E] could
1560 cause memory overwriting.
1561
1562 10. Like Perl, PCRE automatically breaks an unlimited repeat after an empty
1563 string has been matched (to stop an infinite loop). It was not recognizing
1564 a conditional subpattern that could match an empty string if that
1565 subpattern was within another subpattern. For example, it looped when
1566 trying to match (((?(1)X|))*) but it was OK with ((?(1)X|)*) where the
1567 condition was not nested. This bug has been fixed.
1568
1569 12. A pattern like \X?\d or \P{L}?\d in non-UTF-8 mode could cause a backtrack
1570 past the start of the subject in the presence of bytes with the top bit
1571 set, for example "\x8aBCD".
1572
1573 13. Added Perl 5.10 experimental backtracking controls (*FAIL), (*F), (*PRUNE),
1574 (*SKIP), (*THEN), (*COMMIT), and (*ACCEPT).
1575
1576 14. Optimized (?!) to (*FAIL).
1577
1578 15. Updated the test for a valid UTF-8 string to conform to the later RFC 3629.
1579 This restricts code points to be within the range 0 to 0x10FFFF, excluding
1580 the "low surrogate" sequence 0xD800 to 0xDFFF. Previously, PCRE allowed the
1581 full range 0 to 0x7FFFFFFF, as defined by RFC 2279. Internally, it still
1582 does: it's just the validity check that is more restrictive.
1583
1584 16. Inserted checks for integer overflows during escape sequence (backslash)
1585 processing, and also fixed erroneous offset values for syntax errors during
1586 backslash processing.
1587
1588 17. Fixed another case of looking too far back in non-UTF-8 mode (cf 12 above)
1589 for patterns like [\PPP\x8a]{1,}\x80 with the subject "A\x80".
1590
1591 18. An unterminated class in a pattern like (?1)\c[ with a "forward reference"
1592 caused an overrun.
1593
1594 19. A pattern like (?:[\PPa*]*){8,} which had an "extended class" (one with
1595 something other than just ASCII characters) inside a group that had an
1596 unlimited repeat caused a loop at compile time (while checking to see
1597 whether the group could match an empty string).
1598
1599 20. Debugging a pattern containing \p or \P could cause a crash. For example,
1600 [\P{Any}] did so. (Error in the code for printing property names.)
1601
1602 21. An orphan \E inside a character class could cause a crash.
1603
1604 22. A repeated capturing bracket such as (A)? could cause a wild memory
1605 reference during compilation.
1606
1607 23. There are several functions in pcre_compile() that scan along a compiled
1608 expression for various reasons (e.g. to see if it's fixed length for look
1609 behind). There were bugs in these functions when a repeated \p or \P was
1610 present in the pattern. These operators have additional parameters compared
1611 with \d, etc, and these were not being taken into account when moving along
1612 the compiled data. Specifically:
1613
1614 (a) A item such as \p{Yi}{3} in a lookbehind was not treated as fixed
1615 length.
1616
1617 (b) An item such as \pL+ within a repeated group could cause crashes or
1618 loops.
1619
1620 (c) A pattern such as \p{Yi}+(\P{Yi}+)(?1) could give an incorrect
1621 "reference to non-existent subpattern" error.
1622
1623 (d) A pattern like (\P{Yi}{2}\277)? could loop at compile time.
1624
1625 24. A repeated \S or \W in UTF-8 mode could give wrong answers when multibyte
1626 characters were involved (for example /\S{2}/8g with "A\x{a3}BC").
1627
1628 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
1629
1630 26. Patterns such as [\P{Yi}A] which include \p or \P and just one other
1631 character were causing crashes (broken optimization).
1632
1633 27. Patterns such as (\P{Yi}*\277)* (group with possible zero repeat containing
1634 \p or \P) caused a compile-time loop.
1635
1636 28. More problems have arisen in unanchored patterns when CRLF is a valid line
1637 break. For example, the unstudied pattern [\r\n]A does not match the string
1638 "\r\nA" because change 7.0/46 below moves the current point on by two
1639 characters after failing to match at the start. However, the pattern \nA
1640 *does* match, because it doesn't start till \n, and if [\r\n]A is studied,
1641 the same is true. There doesn't seem any very clean way out of this, but
1642 what I have chosen to do makes the common cases work: PCRE now takes note
1643 of whether there can be an explicit match for \r or \n anywhere in the
1644 pattern, and if so, 7.0/46 no longer applies. As part of this change,
1645 there's a new PCRE_INFO_HASCRORLF option for finding out whether a compiled
1646 pattern has explicit CR or LF references.
1647
1648 29. Added (*CR) etc for changing newline setting at start of pattern.
1649
1650
1651 Version 7.2 19-Jun-07
1652 ---------------------
1653
1654 1. If the fr_FR locale cannot be found for test 3, try the "french" locale,
1655 which is apparently normally available under Windows.
1656
1657 2. Re-jig the pcregrep tests with different newline settings in an attempt
1658 to make them independent of the local environment's newline setting.
1659
1660 3. Add code to configure.ac to remove -g from the CFLAGS default settings.
1661
1662 4. Some of the "internals" tests were previously cut out when the link size
1663 was not 2, because the output contained actual offsets. The recent new
1664 "Z" feature of pcretest means that these can be cut out, making the tests
1665 usable with all link sizes.
1666
1667 5. Implemented Stan Switzer's goto replacement for longjmp() when not using
1668 stack recursion. This gives a massive performance boost under BSD, but just
1669 a small improvement under Linux. However, it saves one field in the frame
1670 in all cases.
1671
1672 6. Added more features from the forthcoming Perl 5.10:
1673
1674 (a) (?-n) (where n is a string of digits) is a relative subroutine or
1675 recursion call. It refers to the nth most recently opened parentheses.
1676
1677 (b) (?+n) is also a relative subroutine call; it refers to the nth next
1678 to be opened parentheses.
1679
1680 (c) Conditions that refer to capturing parentheses can be specified
1681 relatively, for example, (?(-2)... or (?(+3)...
1682
1683 (d) \K resets the start of the current match so that everything before
1684 is not part of it.
1685
1686 (e) \k{name} is synonymous with \k<name> and \k'name' (.NET compatible).
1687
1688 (f) \g{name} is another synonym - part of Perl 5.10's unification of
1689 reference syntax.
1690
1691 (g) (?| introduces a group in which the numbering of parentheses in each
1692 alternative starts with the same number.
1693
1694 (h) \h, \H, \v, and \V match horizontal and vertical whitespace.
1695
1696 7. Added two new calls to pcre_fullinfo(): PCRE_INFO_OKPARTIAL and
1697 PCRE_INFO_JCHANGED.
1698
1699 8. A pattern such as (.*(.)?)* caused pcre_exec() to fail by either not
1700 terminating or by crashing. Diagnosed by Viktor Griph; it was in the code
1701 for detecting groups that can match an empty string.
1702
1703 9. A pattern with a very large number of alternatives (more than several
1704 hundred) was running out of internal workspace during the pre-compile
1705 phase, where pcre_compile() figures out how much memory will be needed. A
1706 bit of new cunning has reduced the workspace needed for groups with
1707 alternatives. The 1000-alternative test pattern now uses 12 bytes of
1708 workspace instead of running out of the 4096 that are available.
1709
1710 10. Inserted some missing (unsigned int) casts to get rid of compiler warnings.
1711
1712 11. Applied patch from Google to remove an optimization that didn't quite work.
1713 The report of the bug said:
1714
1715 pcrecpp::RE("a*").FullMatch("aaa") matches, while
1716 pcrecpp::RE("a*?").FullMatch("aaa") does not, and
1717 pcrecpp::RE("a*?\\z").FullMatch("aaa") does again.
1718
1719 12. If \p or \P was used in non-UTF-8 mode on a character greater than 127
1720 it matched the wrong number of bytes.
1721
1722
1723 Version 7.1 24-Apr-07
1724 ---------------------
1725
1726 1. Applied Bob Rossi and Daniel G's patches to convert the build system to one
1727 that is more "standard", making use of automake and other Autotools. There
1728 is some re-arrangement of the files and adjustment of comments consequent
1729 on this.
1730
1731 2. Part of the patch fixed a problem with the pcregrep tests. The test of -r
1732 for recursive directory scanning broke on some systems because the files
1733 are not scanned in any specific order and on different systems the order
1734 was different. A call to "sort" has been inserted into RunGrepTest for the
1735 approprate test as a short-term fix. In the longer term there may be an
1736 alternative.
1737
1738 3. I had an email from Eric Raymond about problems translating some of PCRE's
1739 man pages to HTML (despite the fact that I distribute HTML pages, some
1740 people do their own conversions for various reasons). The problems
1741 concerned the use of low-level troff macros .br and .in. I have therefore
1742 removed all such uses from the man pages (some were redundant, some could
1743 be replaced by .nf/.fi pairs). The 132html script that I use to generate
1744 HTML has been updated to handle .nf/.fi and to complain if it encounters
1745 .br or .in.
1746
1747 4. Updated comments in configure.ac that get placed in config.h.in and also
1748 arranged for config.h to be included in the distribution, with the name
1749 config.h.generic, for the benefit of those who have to compile without
1750 Autotools (compare pcre.h, which is now distributed as pcre.h.generic).
1751
1752 5. Updated the support (such as it is) for Virtual Pascal, thanks to Stefan
1753 Weber: (1) pcre_internal.h was missing some function renames; (2) updated
1754 makevp.bat for the current PCRE, using the additional files
1755 makevp_c.txt, makevp_l.txt, and pcregexp.pas.
1756
1757 6. A Windows user reported a minor discrepancy with test 2, which turned out
1758 to be caused by a trailing space on an input line that had got lost in his
1759 copy. The trailing space was an accident, so I've just removed it.
1760
1761 7. Add -Wl,-R... flags in pcre-config.in for *BSD* systems, as I'm told
1762 that is needed.
1763
1764 8. Mark ucp_table (in ucptable.h) and ucp_gentype (in pcre_ucp_searchfuncs.c)
1765 as "const" (a) because they are and (b) because it helps the PHP
1766 maintainers who have recently made a script to detect big data structures
1767 in the php code that should be moved to the .rodata section. I remembered
1768 to update Builducptable as well, so it won't revert if ucptable.h is ever
1769 re-created.
1770
1771 9. Added some extra #ifdef SUPPORT_UTF8 conditionals into pcretest.c,
1772 pcre_printint.src, pcre_compile.c, pcre_study.c, and pcre_tables.c, in
1773 order to be able to cut out the UTF-8 tables in the latter when UTF-8
1774 support is not required. This saves 1.5-2K of code, which is important in
1775 some applications.
1776
1777 Later: more #ifdefs are needed in pcre_ord2utf8.c and pcre_valid_utf8.c
1778 so as not to refer to the tables, even though these functions will never be
1779 called when UTF-8 support is disabled. Otherwise there are problems with a
1780 shared library.
1781
1782 10. Fixed two bugs in the emulated memmove() function in pcre_internal.h:
1783
1784 (a) It was defining its arguments as char * instead of void *.
1785
1786 (b) It was assuming that all moves were upwards in memory; this was true
1787 a long time ago when I wrote it, but is no longer the case.
1788
1789 The emulated memove() is provided for those environments that have neither
1790 memmove() nor bcopy(). I didn't think anyone used it these days, but that
1791 is clearly not the case, as these two bugs were recently reported.
1792
1793 11. The script PrepareRelease is now distributed: it calls 132html, CleanTxt,
1794 and Detrail to create the HTML documentation, the .txt form of the man
1795 pages, and it removes trailing spaces from listed files. It also creates
1796 pcre.h.generic and config.h.generic from pcre.h and config.h. In the latter
1797 case, it wraps all the #defines with #ifndefs. This script should be run
1798 before "make dist".
1799
1800 12. Fixed two fairly obscure bugs concerned with quantified caseless matching
1801 with Unicode property support.
1802
1803 (a) For a maximizing quantifier, if the two different cases of the
1804 character were of different lengths in their UTF-8 codings (there are
1805 some cases like this - I found 11), and the matching function had to
1806 back up over a mixture of the two cases, it incorrectly assumed they
1807 were both the same length.
1808
1809 (b) When PCRE was configured to use the heap rather than the stack for
1810 recursion during matching, it was not correctly preserving the data for
1811 the other case of a UTF-8 character when checking ahead for a match
1812 while processing a minimizing repeat. If the check also involved
1813 matching a wide character, but failed, corruption could cause an
1814 erroneous result when trying to check for a repeat of the original
1815 character.
1816
1817 13. Some tidying changes to the testing mechanism:
1818
1819 (a) The RunTest script now detects the internal link size and whether there
1820 is UTF-8 and UCP support by running ./pcretest -C instead of relying on
1821 values substituted by "configure". (The RunGrepTest script already did
1822 this for UTF-8.) The configure.ac script no longer substitutes the
1823 relevant variables.
1824
1825 (b) The debugging options /B and /D in pcretest show the compiled bytecode
1826 with length and offset values. This means that the output is different
1827 for different internal link sizes. Test 2 is skipped for link sizes
1828 other than 2 because of this, bypassing the problem. Unfortunately,
1829 there was also a test in test 3 (the locale tests) that used /B and
1830 failed for link sizes other than 2. Rather than cut the whole test out,
1831 I have added a new /Z option to pcretest that replaces the length and
1832 offset values with spaces. This is now used to make test 3 independent
1833 of link size. (Test 2 will be tidied up later.)
1834
1835 14. If erroroffset was passed as NULL to pcre_compile, it provoked a
1836 segmentation fault instead of returning the appropriate error message.
1837
1838 15. In multiline mode when the newline sequence was set to "any", the pattern
1839 ^$ would give a match between the \r and \n of a subject such as "A\r\nB".
1840 This doesn't seem right; it now treats the CRLF combination as the line
1841 ending, and so does not match in that case. It's only a pattern such as ^$
1842 that would hit this one: something like ^ABC$ would have failed after \r
1843 and then tried again after \r\n.
1844
1845 16. Changed the comparison command for RunGrepTest from "diff -u" to "diff -ub"
1846 in an attempt to make files that differ only in their line terminators
1847 compare equal. This works on Linux.
1848
1849 17. Under certain error circumstances pcregrep might try to free random memory
1850 as it exited. This is now fixed, thanks to valgrind.
1851
1852 19. In pcretest, if the pattern /(?m)^$/g<any> was matched against the string
1853 "abc\r\n\r\n", it found an unwanted second match after the second \r. This
1854 was because its rules for how to advance for /g after matching an empty
1855 string at the end of a line did not allow for this case. They now check for
1856 it specially.
1857
1858 20. pcretest is supposed to handle patterns and data of any length, by
1859 extending its buffers when necessary. It was getting this wrong when the
1860 buffer for a data line had to be extended.
1861
1862 21. Added PCRE_NEWLINE_ANYCRLF which is like ANY, but matches only CR, LF, or
1863 CRLF as a newline sequence.
1864
1865 22. Code for handling Unicode properties in pcre_dfa_exec() wasn't being cut
1866 out by #ifdef SUPPORT_UCP. This did no harm, as it could never be used, but
1867 I have nevertheless tidied it up.
1868
1869 23. Added some casts to kill warnings from HP-UX ia64 compiler.
1870
1871 24. Added a man page for pcre-config.
1872
1873
1874 Version 7.0 19-Dec-06
1875 ---------------------
1876
1877 1. Fixed a signed/unsigned compiler warning in pcre_compile.c, shown up by
1878 moving to gcc 4.1.1.
1879
1880 2. The -S option for pcretest uses setrlimit(); I had omitted to #include
1881 sys/time.h, which is documented as needed for this function. It doesn't
1882 seem to matter on Linux, but it showed up on some releases of OS X.
1883
1884 3. It seems that there are systems where bytes whose values are greater than
1885 127 match isprint() in the "C" locale. The "C" locale should be the
1886 default when a C program starts up. In most systems, only ASCII printing
1887 characters match isprint(). This difference caused the output from pcretest
1888 to vary, making some of the tests fail. I have changed pcretest so that:
1889
1890 (a) When it is outputting text in the compiled version of a pattern, bytes
1891 other than 32-126 are always shown as hex escapes.
1892
1893 (b) When it is outputting text that is a matched part of a subject string,
1894 it does the same, unless a different locale has been set for the match
1895 (using the /L modifier). In this case, it uses isprint() to decide.
1896
1897 4. Fixed a major bug that caused incorrect computation of the amount of memory
1898 required for a compiled pattern when options that changed within the
1899 pattern affected the logic of the preliminary scan that determines the
1900 length. The relevant options are -x, and -i in UTF-8 mode. The result was
1901 that the computed length was too small. The symptoms of this bug were
1902 either the PCRE error "internal error: code overflow" from pcre_compile(),
1903 or a glibc crash with a message such as "pcretest: free(): invalid next
1904 size (fast)". Examples of patterns that provoked this bug (shown in
1905 pcretest format) are:
1906
1907 /(?-x: )/x
1908 /(?x)(?-x: \s*#\s*)/
1909 /((?i)[\x{c0}])/8
1910 /(?i:[\x{c0}])/8
1911
1912 HOWEVER: Change 17 below makes this fix obsolete as the memory computation
1913 is now done differently.
1914
1915 5. Applied patches from Google to: (a) add a QuoteMeta function to the C++
1916 wrapper classes; (b) implement a new function in the C++ scanner that is
1917 more efficient than the old way of doing things because it avoids levels of
1918 recursion in the regex matching; (c) add a paragraph to the documentation
1919 for the FullMatch() function.
1920
1921 6. The escape sequence \n was being treated as whatever was defined as
1922 "newline". Not only was this contrary to the documentation, which states
1923 that \n is character 10 (hex 0A), but it also went horribly wrong when
1924 "newline" was defined as CRLF. This has been fixed.
1925
1926 7. In pcre_dfa_exec.c the value of an unsigned integer (the variable called c)
1927 was being set to -1 for the "end of line" case (supposedly a value that no
1928 character can have). Though this value is never used (the check for end of
1929 line is "zero bytes in current character"), it caused compiler complaints.
1930 I've changed it to 0xffffffff.
1931
1932 8. In pcre_version.c, the version string was being built by a sequence of
1933 C macros that, in the event of PCRE_PRERELEASE being defined as an empty
1934 string (as it is for production releases) called a macro with an empty
1935 argument. The C standard says the result of this is undefined. The gcc
1936 compiler treats it as an empty string (which was what was wanted) but it is
1937 reported that Visual C gives an error. The source has been hacked around to
1938 avoid this problem.
1939
1940 9. On the advice of a Windows user, included <io.h> and <fcntl.h> in Windows
1941 builds of pcretest, and changed the call to _setmode() to use _O_BINARY
1942 instead of 0x8000. Made all the #ifdefs test both _WIN32 and WIN32 (not all
1943 of them did).
1944
1945 10. Originally, pcretest opened its input and output without "b"; then I was
1946 told that "b" was needed in some environments, so it was added for release
1947 5.0 to both the input and output. (It makes no difference on Unix-like
1948 systems.) Later I was told that it is wrong for the input on Windows. I've
1949 now abstracted the modes into two macros, to make it easier to fiddle with
1950 them, and removed "b" from the input mode under Windows.
1951
1952 11. Added pkgconfig support for the C++ wrapper library, libpcrecpp.
1953
1954 12. Added -help and --help to pcretest as an official way of being reminded
1955 of the options.
1956
1957 13. Removed some redundant semicolons after macro calls in pcrecpparg.h.in
1958 and pcrecpp.cc because they annoy compilers at high warning levels.
1959
1960 14. A bit of tidying/refactoring in pcre_exec.c in the main bumpalong loop.
1961
1962 15. Fixed an occurrence of == in configure.ac that should have been = (shell
1963 scripts are not C programs :-) and which was not noticed because it works
1964 on Linux.
1965
1966 16. pcretest is supposed to handle any length of pattern and data line (as one
1967 line or as a continued sequence of lines) by extending its input buffer if
1968 necessary. This feature was broken for very long pattern lines, leading to
1969 a string of junk being passed to pcre_compile() if the pattern was longer
1970 than about 50K.
1971
1972 17. I have done a major re-factoring of the way pcre_compile() computes the
1973 amount of memory needed for a compiled pattern. Previously, there was code
1974 that made a preliminary scan of the pattern in order to do this. That was
1975 OK when PCRE was new, but as the facilities have expanded, it has become
1976 harder and harder to keep it in step with the real compile phase, and there
1977 have been a number of bugs (see for example, 4 above). I have now found a
1978 cunning way of running the real compile function in a "fake" mode that
1979 enables it to compute how much memory it would need, while actually only
1980 ever using a few hundred bytes of working memory and without too many
1981 tests of the mode. This should make future maintenance and development
1982 easier. A side effect of this work is that the limit of 200 on the nesting
1983 depth of parentheses has been removed (though this was never a serious
1984 limitation, I suspect). However, there is a downside: pcre_compile() now
1985 runs more slowly than before (30% or more, depending on the pattern). I
1986 hope this isn't a big issue. There is no effect on runtime performance.
1987
1988 18. Fixed a minor bug in pcretest: if a pattern line was not terminated by a
1989 newline (only possible for the last line of a file) and it was a
1990 pattern that set a locale (followed by /Lsomething), pcretest crashed.
1991
1992 19. Added additional timing features to pcretest. (1) The -tm option now times
1993 matching only, not compiling. (2) Both -t and -tm can be followed, as a
1994 separate command line item, by a number that specifies the number of
1995 repeats to use when timing. The default is 50000; this gives better
1996 precision, but takes uncomfortably long for very large patterns.
1997
1998 20. Extended pcre_study() to be more clever in cases where a branch of a
1999 subpattern has no definite first character. For example, (a*|b*)[cd] would
2000 previously give no result from pcre_study(). Now it recognizes that the
2001 first character must be a, b, c, or d.
2002
2003 21. There was an incorrect error "recursive call could loop indefinitely" if
2004 a subpattern (or the entire pattern) that was being tested for matching an
2005 empty string contained only one non-empty item after a nested subpattern.
2006 For example, the pattern (?>\x{100}*)\d(?R) provoked this error
2007 incorrectly, because the \d was being skipped in the check.
2008
2009 22. The pcretest program now has a new pattern option /B and a command line
2010 option -b, which is equivalent to adding /B to every pattern. This causes
2011 it to show the compiled bytecode, without the additional information that
2012 -d shows. The effect of -d is now the same as -b with -i (and similarly, /D
2013 is the same as /B/I).
2014
2015 23. A new optimization is now able automatically to treat some sequences such
2016 as a*b as a*+b. More specifically, if something simple (such as a character
2017 or a simple class like \d) has an unlimited quantifier, and is followed by
2018 something that cannot possibly match the quantified thing, the quantifier
2019 is automatically "possessified".
2020
2021 24. A recursive reference to a subpattern whose number was greater than 39
2022 went wrong under certain circumstances in UTF-8 mode. This bug could also
2023 have affected the operation of pcre_study().
2024
2025 25. Realized that a little bit of performance could be had by replacing
2026 (c & 0xc0) == 0xc0 with c >= 0xc0 when processing UTF-8 characters.
2027
2028 26. Timing data from pcretest is now shown to 4 decimal places instead of 3.
2029
2030 27. Possessive quantifiers such as a++ were previously implemented by turning
2031 them into atomic groups such as ($>a+). Now they have their own opcodes,
2032 which improves performance. This includes the automatically created ones
2033 from 23 above.
2034
2035 28. A pattern such as (?=(\w+))\1: which simulates an atomic group using a
2036 lookahead was broken if it was not anchored. PCRE was mistakenly expecting
2037 the first matched character to be a colon. This applied both to named and
2038 numbered groups.
2039
2040 29. The ucpinternal.h header file was missing its idempotency #ifdef.
2041
2042 30. I was sent a "project" file called libpcre.a.dev which I understand makes
2043 building PCRE on Windows easier, so I have included it in the distribution.
2044
2045 31. There is now a check in pcretest against a ridiculously large number being
2046 returned by pcre_exec() or pcre_dfa_exec(). If this happens in a /g or /G
2047 loop, the loop is abandoned.
2048
2049 32. Forward references to subpatterns in conditions such as (?(2)...) where
2050 subpattern 2 is defined later cause pcre_compile() to search forwards in
2051 the pattern for the relevant set of parentheses. This search went wrong
2052 when there were unescaped parentheses in a character class, parentheses
2053 escaped with \Q...\E, or parentheses in a #-comment in /x mode.
2054
2055 33. "Subroutine" calls and backreferences were previously restricted to
2056 referencing subpatterns earlier in the regex. This restriction has now
2057 been removed.
2058
2059 34. Added a number of extra features that are going to be in Perl 5.10. On the
2060 whole, these are just syntactic alternatives for features that PCRE had
2061 previously implemented using the Python syntax or my own invention. The
2062 other formats are all retained for compatibility.
2063
2064 (a) Named groups can now be defined as (?<name>...) or (?'name'...) as well
2065 as (?P<name>...). The new forms, as well as being in Perl 5.10, are
2066 also .NET compatible.
2067
2068 (b) A recursion or subroutine call to a named group can now be defined as
2069 (?&name) as well as (?P>name).
2070
2071 (c) A backreference to a named group can now be defined as \k<name> or
2072 \k'name' as well as (?P=name). The new forms, as well as being in Perl
2073 5.10, are also .NET compatible.
2074
2075 (d) A conditional reference to a named group can now use the syntax
2076 (?(<name>) or (?('name') as well as (?(name).
2077
2078 (e) A "conditional group" of the form (?(DEFINE)...) can be used to define
2079 groups (named and numbered) that are never evaluated inline, but can be
2080 called as "subroutines" from elsewhere. In effect, the DEFINE condition
2081 is always false. There may be only one alternative in such a group.
2082
2083 (f) A test for recursion can be given as (?(R1).. or (?(R&name)... as well
2084 as the simple (?(R). The condition is true only if the most recent
2085 recursion is that of the given number or name. It does not search out
2086 through the entire recursion stack.
2087
2088 (g) The escape \gN or \g{N} has been added, where N is a positive or
2089 negative number, specifying an absolute or relative reference.
2090
2091 35. Tidied to get rid of some further signed/unsigned compiler warnings and
2092 some "unreachable code" warnings.
2093
2094 36. Updated the Unicode property tables to Unicode version 5.0.0. Amongst other
2095 things, this adds five new scripts.
2096
2097 37. Perl ignores orphaned \E escapes completely. PCRE now does the same.
2098 There were also incompatibilities regarding the handling of \Q..\E inside
2099 character classes, for example with patterns like [\Qa\E-\Qz\E] where the
2100 hyphen was adjacent to \Q or \E. I hope I've cleared all this up now.
2101
2102 38. Like Perl, PCRE detects when an indefinitely repeated parenthesized group
2103 matches an empty string, and forcibly breaks the loop. There were bugs in
2104 this code in non-simple cases. For a pattern such as ^(a()*)* matched
2105 against aaaa the result was just "a" rather than "aaaa", for example. Two
2106 separate and independent bugs (that affected different cases) have been
2107 fixed.
2108
2109 39. Refactored the code to abolish the use of different opcodes for small
2110 capturing bracket numbers. This is a tidy that I avoided doing when I
2111 removed the limit on the number of capturing brackets for 3.5 back in 2001.
2112 The new approach is not only tidier, it makes it possible to reduce the
2113 memory needed to fix the previous bug (38).
2114
2115 40. Implemented PCRE_NEWLINE_ANY to recognize any of the Unicode newline
2116 sequences (http://unicode.org/unicode/reports/tr18/) as "newline" when
2117 processing dot, circumflex, or dollar metacharacters, or #-comments in /x
2118 mode.
2119
2120 41. Add \R to match any Unicode newline sequence, as suggested in the Unicode
2121 report.
2122
2123 42. Applied patch, originally from Ari Pollak, modified by Google, to allow
2124 copy construction and assignment in the C++ wrapper.
2125
2126 43. Updated pcregrep to support "--newline=any". In the process, I fixed a
2127 couple of bugs that could have given wrong results in the "--newline=crlf"
2128 case.
2129
2130 44. Added a number of casts and did some reorganization of signed/unsigned int
2131 variables following suggestions from Dair Grant. Also renamed the variable
2132 "this" as "item" because it is a C++ keyword.
2133
2134 45. Arranged for dftables to add
2135
2136 #include "pcre_internal.h"
2137
2138 to pcre_chartables.c because without it, gcc 4.x may remove the array
2139 definition from the final binary if PCRE is built into a static library and
2140 dead code stripping is activated.
2141
2142 46. For an unanchored pattern, if a match attempt fails at the start of a
2143 newline sequence, and the newline setting is CRLF or ANY, and the next two
2144 characters are CRLF, advance by two characters instead of one.
2145
2146
2147 Version 6.7 04-Jul-06
2148 ---------------------
2149
2150 1. In order to handle tests when input lines are enormously long, pcretest has
2151 been re-factored so that it automatically extends its buffers when
2152 necessary. The code is crude, but this _is_ just a test program. The
2153 default size has been increased from 32K to 50K.
2154
2155 2. The code in pcre_study() was using the value of the re argument before
2156 testing it for NULL. (Of course, in any sensible call of the function, it
2157 won't be NULL.)
2158
2159 3. The memmove() emulation function in pcre_internal.h, which is used on
2160 systems that lack both memmove() and bcopy() - that is, hardly ever -
2161 was missing a "static" storage class specifier.
2162
2163 4. When UTF-8 mode was not set, PCRE looped when compiling certain patterns
2164 containing an extended class (one that cannot be represented by a bitmap
2165 because it contains high-valued characters or Unicode property items, e.g.
2166 [\pZ]). Almost always one would set UTF-8 mode when processing such a
2167 pattern, but PCRE should not loop if you do not (it no longer does).
2168 [Detail: two cases were found: (a) a repeated subpattern containing an
2169 extended class; (b) a recursive reference to a subpattern that followed a
2170 previous extended class. It wasn't skipping over the extended class
2171 correctly when UTF-8 mode was not set.]
2172
2173 5. A negated single-character class was not being recognized as fixed-length
2174 in lookbehind assertions such as (?<=[^f]), leading to an incorrect
2175 compile error "lookbehind assertion is not fixed length".
2176
2177 6. The RunPerlTest auxiliary script was showing an unexpected difference
2178 between PCRE and Perl for UTF-8 tests. It turns out that it is hard to
2179 write a Perl script that can interpret lines of an input file either as
2180 byte characters or as UTF-8, which is what "perltest" was being required to
2181 do for the non-UTF-8 and UTF-8 tests, respectively. Essentially what you
2182 can't do is switch easily at run time between having the "use utf8;" pragma
2183 or not. In the end, I fudged it by using the RunPerlTest script to insert
2184 "use utf8;" explicitly for the UTF-8 tests.
2185
2186 7. In multiline (/m) mode, PCRE was matching ^ after a terminating newline at
2187 the end of the subject string, contrary to the documentation and to what
2188 Perl does. This was true of both matching functions. Now it matches only at
2189 the start of the subject and immediately after *internal* newlines.
2190
2191 8. A call of pcre_fullinfo() from pcretest to get the option bits was passing
2192 a pointer to an int instead of a pointer to an unsigned long int. This
2193 caused problems on 64-bit systems.
2194
2195 9. Applied a patch from the folks at Google to pcrecpp.cc, to fix "another
2196 instance of the 'standard' template library not being so standard".
2197
2198 10. There was no check on the number of named subpatterns nor the maximum
2199 length of a subpattern name. The product of these values is used to compute
2200 the size of the memory block for a compiled pattern. By supplying a very
2201 long subpattern name and a large number of named subpatterns, the size
2202 computation could be caused to overflow. This is now prevented by limiting
2203 the length of names to 32 characters, and the number of named subpatterns
2204 to 10,000.
2205
2206 11. Subpatterns that are repeated with specific counts have to be replicated in
2207 the compiled pattern. The size of memory for this was computed from the
2208 length of the subpattern and the repeat count. The latter is limited to
2209 65535, but there was no limit on the former, meaning that integer overflow
2210 could in principle occur. The compiled length of a repeated subpattern is
2211 now limited to 30,000 bytes in order to prevent this.
2212
2213 12. Added the optional facility to have named substrings with the same name.
2214
2215 13. Added the ability to use a named substring as a condition, using the
2216 Python syntax: (?(name)yes|no). This overloads (?(R)... and names that
2217 are numbers (not recommended). Forward references are permitted.
2218
2219 14. Added forward references in named backreferences (if you see what I mean).
2220
2221 15. In UTF-8 mode, with the PCRE_DOTALL option set, a quantified dot in the
2222 pattern could run off the end of the subject. For example, the pattern
2223 "(?s)(.{1,5})"8 did this with the subject "ab".
2224
2225 16. If PCRE_DOTALL or PCRE_MULTILINE were set, pcre_dfa_exec() behaved as if
2226 PCRE_CASELESS was set when matching characters that were quantified with ?
2227 or *.
2228
2229 17. A character class other than a single negated character that had a minimum
2230 but no maximum quantifier - for example [ab]{6,} - was not handled
2231 correctly by pce_dfa_exec(). It would match only one character.
2232
2233 18. A valid (though odd) pattern that looked like a POSIX character
2234 class but used an invalid character after [ (for example [[,abc,]]) caused
2235 pcre_compile() to give the error "Failed: internal error: code overflow" or
2236 in some cases to crash with a glibc free() error. This could even happen if
2237 the pattern terminated after [[ but there just happened to be a sequence of
2238 letters, a binary zero, and a closing ] in the memory that followed.
2239
2240 19. Perl's treatment of octal escapes in the range \400 to \777 has changed
2241 over the years. Originally (before any Unicode support), just the bottom 8
2242 bits were taken. Thus, for example, \500 really meant \100. Nowadays the
2243 output from "man perlunicode" includes this:
2244
2245 The regular expression compiler produces polymorphic opcodes. That
2246 is, the pattern adapts to the data and automatically switches to
2247 the Unicode character scheme when presented with Unicode data--or
2248 instead uses a traditional byte scheme when presented with byte
2249 data.
2250
2251 Sadly, a wide octal escape does not cause a switch, and in a string with
2252 no other multibyte characters, these octal escapes are treated as before.
2253 Thus, in Perl, the pattern /\500/ actually matches \100 but the pattern
2254 /\500|\x{1ff}/ matches \500 or \777 because the whole thing is treated as a
2255 Unicode string.
2256
2257 I have not perpetrated such confusion in PCRE. Up till now, it took just
2258 the bottom 8 bits, as in old Perl. I have now made octal escapes with
2259 values greater than \377 illegal in non-UTF-8 mode. In UTF-8 mode they
2260 translate to the appropriate multibyte character.
2261
2262 29. Applied some refactoring to reduce the number of warnings from Microsoft
2263 and Borland compilers. This has included removing the fudge introduced
2264 seven years ago for the OS/2 compiler (see 2.02/2 below) because it caused
2265 a warning about an unused variable.
2266
2267 21. PCRE has not included VT (character 0x0b) in the set of whitespace
2268 characters since release 4.0, because Perl (from release 5.004) does not.
2269 [Or at least, is documented not to: some releases seem to be in conflict
2270 with the documentation.] However, when a pattern was studied with
2271 pcre_study() and all its branches started with \s, PCRE still included VT
2272 as a possible starting character. Of course, this did no harm; it just
2273 caused an unnecessary match attempt.
2274
2275 22. Removed a now-redundant internal flag bit that recorded the fact that case
2276 dependency changed within the pattern. This was once needed for "required
2277 byte" processing, but is no longer used. This recovers a now-scarce options
2278 bit. Also moved the least significant internal flag bit to the most-
2279 significant bit of the word, which was not previously used (hangover from
2280 the days when it was an int rather than a uint) to free up another bit for
2281 the future.
2282
2283 23. Added support for CRLF line endings as well as CR and LF. As well as the
2284 default being selectable at build time, it can now be changed at runtime
2285 via the PCRE_NEWLINE_xxx flags. There are now options for pcregrep to
2286 specify that it is scanning data with non-default line endings.
2287
2288 24. Changed the definition of CXXLINK to make it agree with the definition of
2289 LINK in the Makefile, by replacing LDFLAGS to CXXFLAGS.
2290
2291 25. Applied Ian Taylor's patches to avoid using another stack frame for tail
2292 recursions. This makes a big different to stack usage for some patterns.
2293
2294 26. If a subpattern containing a named recursion or subroutine reference such
2295 as (?P>B) was quantified, for example (xxx(?P>B)){3}, the calculation of
2296 the space required for the compiled pattern went wrong and gave too small a
2297 value. Depending on the environment, this could lead to "Failed: internal
2298 error: code overflow at offset 49" or "glibc detected double free or
2299 corruption" errors.
2300
2301 27. Applied patches from Google (a) to support the new newline modes and (b) to
2302 advance over multibyte UTF-8 characters in GlobalReplace.
2303
2304 28. Change free() to pcre_free() in pcredemo.c. Apparently this makes a
2305 difference for some implementation of PCRE in some Windows version.
2306
2307 29. Added some extra testing facilities to pcretest:
2308
2309 \q<number> in a data line sets the "match limit" value
2310 \Q<number> in a data line sets the "match recursion limt" value
2311 -S <number> sets the stack size, where <number> is in megabytes
2312
2313 The -S option isn't available for Windows.
2314
2315
2316 Version 6.6 06-Feb-06
2317 ---------------------
2318
2319 1. Change 16(a) for 6.5 broke things, because PCRE_DATA_SCOPE was not defined
2320 in pcreposix.h. I have copied the definition from pcre.h.
2321
2322 2. Change 25 for 6.5 broke compilation in a build directory out-of-tree
2323 because pcre.h is no longer a built file.
2324
2325 3. Added Jeff Friedl's additional debugging patches to pcregrep. These are
2326 not normally included in the compiled code.
2327
2328
2329 Version 6.5 01-Feb-06
2330 ---------------------
2331
2332 1. When using the partial match feature with pcre_dfa_exec(), it was not
2333 anchoring the second and subsequent partial matches at the new starting
2334 point. This could lead to incorrect results. For example, with the pattern
2335 /1234/, partially matching against "123" and then "a4" gave a match.
2336
2337 2. Changes to pcregrep:
2338
2339 (a) All non-match returns from pcre_exec() were being treated as failures
2340 to match the line. Now, unless the error is PCRE_ERROR_NOMATCH, an
2341 error message is output. Some extra information is given for the
2342 PCRE_ERROR_MATCHLIMIT and PCRE_ERROR_RECURSIONLIMIT errors, which are
2343 probably the only errors that are likely to be caused by users (by
2344 specifying a regex that has nested indefinite repeats, for instance).
2345 If there are more than 20 of these errors, pcregrep is abandoned.
2346
2347 (b) A binary zero was treated as data while matching, but terminated the
2348 output line if it was written out. This has been fixed: binary zeroes
2349 are now no different to any other data bytes.
2350
2351 (c) Whichever of the LC_ALL or LC_CTYPE environment variables is set is
2352 used to set a locale for matching. The --locale=xxxx long option has
2353 been added (no short equivalent) to specify a locale explicitly on the
2354 pcregrep command, overriding the environment variables.
2355
2356 (d) When -B was used with -n, some line numbers in the output were one less
2357 than they should have been.
2358
2359 (e) Added the -o (--only-matching) option.
2360
2361 (f) If -A or -C was used with -c (count only), some lines of context were
2362 accidentally printed for the final match.
2363
2364 (g) Added the -H (--with-filename) option.
2365
2366 (h) The combination of options -rh failed to suppress file names for files
2367 that were found from directory arguments.
2368
2369 (i) Added the -D (--devices) and -d (--directories) options.
2370
2371 (j) Added the -F (--fixed-strings) option.
2372
2373 (k) Allow "-" to be used as a file name for -f as well as for a data file.
2374
2375 (l) Added the --colo(u)r option.
2376
2377 (m) Added Jeffrey Friedl's -S testing option, but within #ifdefs so that it
2378 is not present by default.
2379
2380 3. A nasty bug was discovered in the handling of recursive patterns, that is,
2381 items such as (?R) or (?1), when the recursion could match a number of
2382 alternatives. If it matched one of the alternatives, but subsequently,
2383 outside the recursion, there was a failure, the code tried to back up into
2384 the recursion. However, because of the way PCRE is implemented, this is not
2385 possible, and the result was an incorrect result from the match.
2386
2387 In order to prevent this happening, the specification of recursion has
2388 been changed so that all such subpatterns are automatically treated as
2389 atomic groups. Thus, for example, (?R) is treated as if it were (?>(?R)).
2390
2391 4. I had overlooked the fact that, in some locales, there are characters for
2392 which isalpha() is true but neither isupper() nor islower() are true. In
2393 the fr_FR locale, for instance, the \xAA and \xBA characters (ordmasculine
2394 and ordfeminine) are like this. This affected the treatment of \w and \W
2395 when they appeared in character classes, but not when they appeared outside
2396 a character class. The bit map for "word" characters is now created
2397 separately from the results of isalnum() instead of just taking it from the
2398 upper, lower, and digit maps. (Plus the underscore character, of course.)
2399
2400 5. The above bug also affected the handling of POSIX character classes such as
2401 [[:alpha:]] and [[:alnum:]]. These do not have their own bit maps in PCRE's
2402 permanent tables. Instead, the bit maps for such a class were previously
2403 created as the appropriate unions of the upper, lower, and digit bitmaps.
2404 Now they are created by subtraction from the [[:word:]] class, which has
2405 its own bitmap.
2406
2407 6. The [[:blank:]] character class matches horizontal, but not vertical space.
2408 It is created by subtracting the vertical space characters (\x09, \x0a,
2409 \x0b, \x0c) from the [[:space:]] bitmap. Previously, however, the
2410 subtraction was done in the overall bitmap for a character class, meaning
2411 that a class such as [\x0c[:blank:]] was incorrect because \x0c would not
2412 be recognized. This bug has been fixed.
2413
2414 7. Patches from the folks at Google:
2415
2416 (a) pcrecpp.cc: "to handle a corner case that may or may not happen in
2417 real life, but is still worth protecting against".
2418
2419 (b) pcrecpp.cc: "corrects a bug when negative radixes are used with
2420 regular expressions".
2421
2422 (c) pcre_scanner.cc: avoid use of std::count() because not all systems
2423 have it.
2424
2425 (d) Split off pcrecpparg.h from pcrecpp.h and had the former built by
2426 "configure" and the latter not, in order to fix a problem somebody had
2427 with compiling the Arg class on HP-UX.
2428
2429 (e) Improve the error-handling of the C++ wrapper a little bit.
2430
2431 (f) New tests for checking recursion limiting.
2432
2433 8. The pcre_memmove() function, which is used only if the environment does not
2434 have a standard memmove() function (and is therefore rarely compiled),
2435 contained two bugs: (a) use of int instead of size_t, and (b) it was not
2436 returning a result (though PCRE never actually uses the result).
2437
2438 9. In the POSIX regexec() interface, if nmatch is specified as a ridiculously
2439 large number - greater than INT_MAX/(3*sizeof(int)) - REG_ESPACE is
2440 returned instead of calling malloc() with an overflowing number that would
2441 most likely cause subsequent chaos.
2442
2443 10. The debugging option of pcretest was not showing the NO_AUTO_CAPTURE flag.
2444
2445 11. The POSIX flag REG_NOSUB is now supported. When a pattern that was compiled
2446 with this option is matched, the nmatch and pmatch options of regexec() are
2447 ignored.
2448
2449 12. Added REG_UTF8 to the POSIX interface. This is not defined by POSIX, but is
2450 provided in case anyone wants to the the POSIX interface with UTF-8
2451 strings.
2452
2453 13. Added CXXLDFLAGS to the Makefile parameters to provide settings only on the
2454 C++ linking (needed for some HP-UX environments).
2455
2456 14. Avoid compiler warnings in get_ucpname() when compiled without UCP support
2457 (unused parameter) and in the pcre_printint() function (omitted "default"
2458 switch label when the default is to do nothing).
2459
2460 15. Added some code to make it possible, when PCRE is compiled as a C++
2461 library, to replace subject pointers for pcre_exec() with a smart pointer
2462 class, thus making it possible to process discontinuous strings.
2463
2464 16. The two macros PCRE_EXPORT and PCRE_DATA_SCOPE are confusing, and perform
2465 much the same function. They were added by different people who were trying
2466 to make PCRE easy to compile on non-Unix systems. It has been suggested
2467 that PCRE_EXPORT be abolished now that there is more automatic apparatus
2468 for compiling on Windows systems. I have therefore replaced it with
2469 PCRE_DATA_SCOPE. This is set automatically for Windows; if not set it
2470 defaults to "extern" for C or "extern C" for C++, which works fine on
2471 Unix-like systems. It is now possible to override the value of PCRE_DATA_
2472 SCOPE with something explicit in config.h. In addition:
2473
2474 (a) pcreposix.h still had just "extern" instead of either of these macros;
2475 I have replaced it with PCRE_DATA_SCOPE.
2476
2477 (b) Functions such as _pcre_xclass(), which are internal to the library,
2478 but external in the C sense, all had PCRE_EXPORT in their definitions.
2479 This is apparently wrong for the Windows case, so I have removed it.
2480 (It makes no difference on Unix-like systems.)
2481
2482 17. Added a new limit, MATCH_LIMIT_RECURSION, which limits the depth of nesting
2483 of recursive calls to match(). This is different to MATCH_LIMIT because
2484 that limits the total number of calls to match(), not all of which increase
2485 the depth of recursion. Limiting the recursion depth limits the amount of
2486 stack (or heap if NO_RECURSE is set) that is used. The default can be set
2487 when PCRE is compiled, and changed at run time. A patch from Google adds
2488 this functionality to the C++ interface.
2489
2490 18. Changes to the handling of Unicode character properties:
2491
2492 (a) Updated the table to Unicode 4.1.0.
2493
2494 (b) Recognize characters that are not in the table as "Cn" (undefined).
2495
2496 (c) I revised the way the table is implemented to a much improved format
2497 which includes recognition of ranges. It now supports the ranges that
2498 are defined in UnicodeData.txt, and it also amalgamates other
2499 characters into ranges. This has reduced the number of entries in the
2500 table from around 16,000 to around 3,000, thus reducing its size
2501 considerably. I realized I did not need to use a tree structure after
2502 all - a binary chop search is just as efficient. Having reduced the
2503 number of entries, I extended their size from 6 bytes to 8 bytes to
2504 allow for more data.
2505
2506 (d) Added support for Unicode script names via properties such as \p{Han}.
2507
2508 19. In UTF-8 mode, a backslash followed by a non-Ascii character was not
2509 matching that character.
2510
2511 20. When matching a repeated Unicode property with a minimum greater than zero,
2512 (for example \pL{2,}), PCRE could look past the end of the subject if it
2513 reached it while seeking the minimum number of characters. This could
2514 happen only if some of the characters were more than one byte long, because
2515 there is a check for at least the minimum number of bytes.
2516
2517 21. Refactored the implementation of \p and \P so as to be more general, to
2518 allow for more different types of property in future. This has changed the
2519 compiled form incompatibly. Anybody with saved compiled patterns that use
2520 \p or \P will have to recompile them.
2521
2522 22. Added "Any" and "L&" to the supported property types.
2523
2524 23. Recognize \x{...} as a code point specifier, even when not in UTF-8 mode,
2525 but give a compile time error if the value is greater than 0xff.
2526
2527 24. The man pages for pcrepartial, pcreprecompile, and pcre_compile2 were
2528 accidentally not being installed or uninstalled.
2529
2530 25. The pcre.h file was built from pcre.h.in, but the only changes that were
2531 made were to insert the current release number. This seemed silly, because
2532 it made things harder for people building PCRE on systems that don't run
2533 "configure". I have turned pcre.h into a distributed file, no longer built
2534 by "configure", with the version identification directly included. There is
2535 no longer a pcre.h.in file.
2536
2537 However, this change necessitated a change to the pcre-config script as
2538 well. It is built from pcre-config.in, and one of the substitutions was the
2539 release number. I have updated configure.ac so that ./configure now finds
2540 the release number by grepping pcre.h.
2541
2542 26. Added the ability to run the tests under valgrind.
2543
2544
2545 Version 6.4 05-Sep-05
2546 ---------------------
2547
2548 1. Change 6.0/10/(l) to pcregrep introduced a bug that caused separator lines
2549 "--" to be printed when multiple files were scanned, even when none of the
2550 -A, -B, or -C options were used. This is not compatible with Gnu grep, so I
2551 consider it to be a bug, and have restored the previous behaviour.
2552
2553 2. A couple of code tidies to get rid of compiler warnings.
2554
2555 3. The pcretest program used to cheat by referring to symbols in the library
2556 whose names begin with _pcre_. These are internal symbols that are not
2557 really supposed to be visible externally, and in some environments it is
2558 possible to suppress them. The cheating is now confined to including
2559 certain files from the library's source, which is a bit cleaner.
2560
2561 4. Renamed pcre.in as pcre.h.in to go with pcrecpp.h.in; it also makes the
2562 file's purpose clearer.
2563
2564 5. Reorganized pcre_ucp_findchar().
2565
2566
2567 Version 6.3 15-Aug-05
2568 ---------------------
2569
2570 1. The file libpcre.pc.in did not have general read permission in the tarball.
2571
2572 2. There were some problems when building without C++ support:
2573
2574 (a) If C++ support was not built, "make install" and "make test" still
2575 tried to test it.
2576
2577 (b) There were problems when the value of CXX was explicitly set. Some
2578 changes have been made to try to fix these, and ...
2579
2580 (c) --disable-cpp can now be used to explicitly disable C++ support.
2581
2582 (d) The use of @CPP_OBJ@ directly caused a blank line preceded by a
2583 backslash in a target when C++ was disabled. This confuses some
2584 versions of "make", apparently. Using an intermediate variable solves
2585 this. (Same for CPP_LOBJ.)
2586
2587 3. $(LINK_FOR_BUILD) now includes $(CFLAGS_FOR_BUILD) and $(LINK)
2588 (non-Windows) now includes $(CFLAGS) because these flags are sometimes
2589 necessary on certain architectures.
2590
2591 4. Added a setting of -export-symbols-regex to the link command to remove
2592 those symbols that are exported in the C sense, but actually are local
2593 within the library, and not documented. Their names all begin with
2594 "_pcre_". This is not a perfect job, because (a) we have to except some
2595 symbols that pcretest ("illegally") uses, and (b) the facility isn't always
2596 available (and never for static libraries). I have made a note to try to
2597 find a way round (a) in the future.
2598
2599
2600 Version 6.2 01-Aug-05
2601 ---------------------
2602
2603 1. There was no test for integer overflow of quantifier values. A construction
2604 such as {1111111111111111} would give undefined results. What is worse, if
2605 a minimum quantifier for a parenthesized subpattern overflowed and became
2606 negative, the calculation of the memory size went wrong. This could have
2607 led to memory overwriting.
2608
2609 2. Building PCRE using VPATH was broken. Hopefully it is now fixed.
2610
2611 3. Added "b" to the 2nd argument of fopen() in dftables.c, for non-Unix-like
2612 operating environments where this matters.
2613
2614 4. Applied Giuseppe Maxia's patch to add additional features for controlling
2615 PCRE options from within the C++ wrapper.
2616
2617 5. Named capturing subpatterns were not being correctly counted when a pattern
2618 was compiled. This caused two problems: (a) If there were more than 100
2619 such subpatterns, the calculation of the memory needed for the whole
2620 compiled pattern went wrong, leading to an overflow error. (b) Numerical
2621 back references of the form \12, where the number was greater than 9, were
2622 not recognized as back references, even though there were sufficient
2623 previous subpatterns.
2624
2625 6. Two minor patches to pcrecpp.cc in order to allow it to compile on older
2626 versions of gcc, e.g. 2.95.4.
2627
2628
2629 Version 6.1 21-Jun-05
2630 ---------------------
2631
2632 1. There was one reference to the variable "posix" in pcretest.c that was not
2633 surrounded by "#if !defined NOPOSIX".
2634
2635 2. Make it possible to compile pcretest without DFA support, UTF8 support, or
2636 the cross-check on the old pcre_info() function, for the benefit of the
2637 cut-down version of PCRE that is currently imported into Exim.
2638
2639 3. A (silly) pattern starting with (?i)(?-i) caused an internal space
2640 allocation error. I've done the easy fix, which wastes 2 bytes for sensible
2641 patterns that start (?i) but I don't think that matters. The use of (?i) is
2642 just an example; this all applies to the other options as well.
2643
2644 4. Since libtool seems to echo the compile commands it is issuing, the output
2645 from "make" can be reduced a bit by putting "@" in front of each libtool
2646 compile command.
2647
2648 5. Patch from the folks at Google for configure.in to be a bit more thorough
2649 in checking for a suitable C++ installation before trying to compile the
2650 C++ stuff. This should fix a reported problem when a compiler was present,
2651 but no suitable headers.
2652
2653 6. The man pages all had just "PCRE" as their title. I have changed them to
2654 be the relevant file name. I have also arranged that these names are
2655 retained in the file doc/pcre.txt, which is a concatenation in text format
2656 of all the man pages except the little individual ones for each function.
2657
2658 7. The NON-UNIX-USE file had not been updated for the different set of source
2659 files that come with release 6. I also added a few comments about the C++
2660 wrapper.
2661
2662
2663 Version 6.0 07-Jun-05
2664 ---------------------
2665
2666 1. Some minor internal re-organization to help with my DFA experiments.
2667
2668 2. Some missing #ifdef SUPPORT_UCP conditionals in pcretest and printint that
2669 didn't matter for the library itself when fully configured, but did matter
2670 when compiling without UCP support, or within Exim, where the ucp files are
2671 not imported.
2672
2673 3. Refactoring of the library code to split up the various functions into
2674 different source modules. The addition of the new DFA matching code (see
2675 below) to a single monolithic source would have made it really too
2676 unwieldy, quite apart from causing all the code to be include in a
2677 statically linked application, when only some functions are used. This is
2678 relevant even without the DFA addition now that patterns can be compiled in
2679 one application and matched in another.
2680
2681 The downside of splitting up is that there have to be some external
2682 functions and data tables that are used internally in different modules of
2683 the library but which are not part of the API. These have all had their
2684 names changed to start with "_pcre_" so that they are unlikely to clash
2685 with other external names.
2686
2687 4. Added an alternate matching function, pcre_dfa_exec(), which matches using
2688 a different (DFA) algorithm. Although it is slower than the original
2689 function, it does have some advantages for certain types of matching
2690 problem.
2691
2692 5. Upgrades to pcretest in order to test the features of pcre_dfa_exec(),
2693 including restarting after a partial match.
2694
2695 6. A patch for pcregrep that defines INVALID_FILE_ATTRIBUTES if it is not
2696 defined when compiling for Windows was sent to me. I have put it into the
2697 code, though I have no means of testing or verifying it.
2698
2699 7. Added the pcre_refcount() auxiliary function.
2700
2701 8. Added the PCRE_FIRSTLINE option. This constrains an unanchored pattern to
2702 match before or at the first newline in the subject string. In pcretest,
2703 the /f option on a pattern can be used to set this.
2704
2705 9. A repeated \w when used in UTF-8 mode with characters greater than 256
2706 would behave wrongly. This has been present in PCRE since release 4.0.
2707
2708 10. A number of changes to the pcregrep command:
2709
2710 (a) Refactored how -x works; insert ^(...)$ instead of setting
2711 PCRE_ANCHORED and checking the length, in preparation for adding
2712 something similar for -w.
2713
2714 (b) Added the -w (match as a word) option.
2715
2716 (c) Refactored the way lines are read and buffered so as to have more
2717 than one at a time available.
2718
2719 (d) Implemented a pcregrep test script.
2720
2721 (e) Added the -M (multiline match) option. This allows patterns to match
2722 over several lines of the subject. The buffering ensures that at least
2723 8K, or the rest of the document (whichever is the shorter) is available
2724 for matching (and similarly the previous 8K for lookbehind assertions).
2725
2726 (f) Changed the --help output so that it now says
2727
2728 -w, --word-regex(p)
2729
2730 instead of two lines, one with "regex" and the other with "regexp"
2731 because that confused at least one person since the short forms are the
2732 same. (This required a bit of code, as the output is generated
2733 automatically from a table. It wasn't just a text change.)
2734
2735 (g) -- can be used to terminate pcregrep options if the next thing isn't an
2736 option but starts with a hyphen. Could be a pattern or a path name
2737 starting with a hyphen, for instance.
2738
2739 (h) "-" can be given as a file name to represent stdin.
2740
2741 (i) When file names are being printed, "(standard input)" is used for
2742 the standard input, for compatibility with GNU grep. Previously
2743 "<stdin>" was used.
2744
2745 (j) The option --label=xxx can be used to supply a name to be used for
2746 stdin when file names are being printed. There is no short form.
2747
2748 (k) Re-factored the options decoding logic because we are going to add
2749 two more options that take data. Such options can now be given in four
2750 different ways, e.g. "-fname", "-f name", "--file=name", "--file name".
2751
2752 (l) Added the -A, -B, and -C options for requesting that lines of context
2753 around matches be printed.
2754
2755 (m) Added the -L option to print the names of files that do not contain
2756 any matching lines, that is, the complement of -l.
2757
2758 (n) The return code is 2 if any file cannot be opened, but pcregrep does
2759 continue to scan other files.
2760
2761 (o) The -s option was incorrectly implemented. For compatibility with other
2762 greps, it now suppresses the error message for a non-existent or non-
2763 accessible file (but not the return code). There is a new option called
2764 -q that suppresses the output of matching lines, which was what -s was
2765 previously doing.
2766
2767 (p) Added --include and --exclude options to specify files for inclusion
2768 and exclusion when recursing.
2769
2770 11. The Makefile was not using the Autoconf-supported LDFLAGS macro properly.
2771 Hopefully, it now does.
2772
2773 12. Missing cast in pcre_study().
2774
2775 13. Added an "uninstall" target to the makefile.
2776
2777 14. Replaced "extern" in the function prototypes in Makefile.in with
2778 "PCRE_DATA_SCOPE", which defaults to 'extern' or 'extern "C"' in the Unix
2779 world, but is set differently for Windows.
2780
2781 15. Added a second compiling function called pcre_compile2(). The only
2782 difference is that it has an extra argument, which is a pointer to an
2783 integer error code. When there is a compile-time failure, this is set
2784 non-zero, in addition to the error test pointer being set to point to an
2785 error message. The new argument may be NULL if no error number is required
2786 (but then you may as well call pcre_compile(), which is now just a
2787 wrapper). This facility is provided because some applications need a
2788 numeric error indication, but it has also enabled me to tidy up the way
2789 compile-time errors are handled in the POSIX wrapper.
2790
2791 16. Added VPATH=.libs to the makefile; this should help when building with one
2792 prefix path and installing with another. (Or so I'm told by someone who
2793 knows more about this stuff than I do.)
2794
2795 17. Added a new option, REG_DOTALL, to the POSIX function regcomp(). This
2796 passes PCRE_DOTALL to the pcre_compile() function, making the "." character
2797 match everything, including newlines. This is not POSIX-compatible, but
2798 somebody wanted the feature. From pcretest it can be activated by using
2799 both the P and the s flags.
2800
2801 18. AC_PROG_LIBTOOL appeared twice in Makefile.in. Removed one.
2802
2803 19. libpcre.pc was being incorrectly installed as executable.
2804
2805 20. A couple of places in pcretest check for end-of-line by looking for '\n';
2806 it now also looks for '\r' so that it will work unmodified on Windows.
2807
2808 21. Added Google's contributed C++ wrapper to the distribution.
2809
2810 22. Added some untidy missing memory free() calls in pcretest, to keep
2811 Electric Fence happy when testing.
2812
2813
2814
2815 Version 5.0 13-Sep-04
2816 ---------------------
2817
2818 1. Internal change: literal characters are no longer packed up into items
2819 containing multiple characters in a single byte-string. Each character
2820 is now matched using a separate opcode. However, there may be more than one
2821 byte in the character in UTF-8 mode.
2822
2823 2. The pcre_callout_block structure has two new fields: pattern_position and
2824 next_item_length. These contain the offset in the pattern to the next match
2825 item, and its length, respectively.
2826
2827 3. The PCRE_AUTO_CALLOUT option for pcre_compile() requests the automatic
2828 insertion of callouts before each pattern item. Added the /C option to
2829 pcretest to make use of this.
2830
2831 4. On the advice of a Windows user, the lines
2832
2833 #if defined(_WIN32) || defined(WIN32)
2834 _setmode( _fileno( stdout ), 0x8000 );
2835 #endif /* defined(_WIN32) || defined(WIN32) */
2836
2837 have been added to the source of pcretest. This apparently does useful
2838 magic in relation to line terminators.
2839
2840 5. Changed "r" and "w" in the calls to fopen() in pcretest to "rb" and "wb"
2841 for the benefit of those environments where the "b" makes a difference.
2842
2843 6. The icc compiler has the same options as gcc, but "configure" doesn't seem
2844 to know about it. I have put a hack into configure.in that adds in code
2845 to set GCC=yes if CC=icc. This seems to end up at a point in the
2846 generated configure script that is early enough to affect the setting of
2847 compiler options, which is what is needed, but I have no means of testing
2848 whether it really works. (The user who reported this had patched the
2849 generated configure script, which of course I cannot do.)
2850
2851 LATER: After change 22 below (new libtool files), the configure script
2852 seems to know about icc (and also ecc). Therefore, I have commented out
2853 this hack in configure.in.
2854
2855 7. Added support for pkg-config (2 patches were sent in).
2856
2857 8. Negated POSIX character classes that used a combination of internal tables
2858 were completely broken. These were [[:^alpha:]], [[:^alnum:]], and
2859 [[:^ascii]]. Typically, they would match almost any characters. The other
2860 POSIX classes were not broken in this way.
2861
2862 9. Matching the pattern "\b.*?" against "ab cd", starting at offset 1, failed
2863 to find the match, as PCRE was deluded into thinking that the match had to
2864 start at the start point or following a newline. The same bug applied to
2865 patterns with negative forward assertions or any backward assertions
2866 preceding ".*" at the start, unless the pattern required a fixed first
2867 character. This was a failing pattern: "(?!.bcd).*". The bug is now fixed.
2868
2869 10. In UTF-8 mode, when moving forwards in the subject after a failed match
2870 starting at the last subject character, bytes beyond the end of the subject
2871 string were read.
2872
2873 11. Renamed the variable "class" as "classbits" to make life easier for C++
2874 users. (Previously there was a macro definition, but it apparently wasn't
2875 enough.)
2876
2877 12. Added the new field "tables" to the extra data so that tables can be passed
2878 in at exec time, or the internal tables can be re-selected. This allows
2879 a compiled regex to be saved and re-used at a later time by a different
2880 program that might have everything at different addresses.
2881
2882 13. Modified the pcre-config script so that, when run on Solaris, it shows a
2883 -R library as well as a -L library.
2884
2885 14. The debugging options of pcretest (-d on the command line or D on a
2886 pattern) showed incorrect output for anything following an extended class
2887 that contained multibyte characters and which was followed by a quantifier.
2888
2889 15. Added optional support for general category Unicode character properties
2890 via the \p, \P, and \X escapes. Unicode property support implies UTF-8
2891 support. It adds about 90K to the size of the library. The meanings of the
2892 inbuilt class escapes such as \d and \s have NOT been changed.
2893
2894 16. Updated pcredemo.c to include calls to free() to release the memory for the
2895 compiled pattern.
2896
2897 17. The generated file chartables.c was being created in the source directory
2898 instead of in the building directory. This caused the build to fail if the
2899 source directory was different from the building directory, and was
2900 read-only.
2901
2902 18. Added some sample Win commands from Mark Tetrode into the NON-UNIX-USE
2903 file. No doubt somebody will tell me if they don't make sense... Also added
2904 Dan Mooney's comments about building on OpenVMS.
2905
2906 19. Added support for partial matching via the PCRE_PARTIAL option for
2907 pcre_exec() and the \P data escape in pcretest.
2908
2909 20. Extended pcretest with 3 new pattern features:
2910
2911 (i) A pattern option of the form ">rest-of-line" causes pcretest to
2912 write the compiled pattern to the file whose name is "rest-of-line".
2913 This is a straight binary dump of the data, with the saved pointer to
2914 the character tables forced to be NULL. The study data, if any, is
2915 written too. After writing, pcretest reads a new pattern.
2916
2917 (ii) If, instead of a pattern, "<rest-of-line" is given, pcretest reads a
2918 compiled pattern from the given file. There must not be any
2919 occurrences of "<" in the file name (pretty unlikely); if there are,
2920 pcretest will instead treat the initial "<" as a pattern delimiter.
2921 After reading in the pattern, pcretest goes on to read data lines as
2922 usual.
2923
2924 (iii) The F pattern option causes pcretest to flip the bytes in the 32-bit
2925 and 16-bit fields in a compiled pattern, to simulate a pattern that
2926 was compiled on a host of opposite endianness.
2927
2928 21. The pcre-exec() function can now cope with patterns that were compiled on
2929 hosts of opposite endianness, with this restriction:
2930
2931 As for any compiled expression that is saved and used later, the tables
2932 pointer field cannot be preserved; the extra_data field in the arguments
2933 to pcre_exec() should be used to pass in a tables address if a value
2934 other than the default internal tables were used at compile time.
2935
2936 22. Calling pcre_exec() with a negative value of the "ovecsize" parameter is
2937 now diagnosed as an error. Previously, most of the time, a negative number
2938 would have been treated as zero, but if in addition "ovector" was passed as
2939 NULL, a crash could occur.
2940
2941 23. Updated the files ltmain.sh, config.sub, config.guess, and aclocal.m4 with
2942 new versions from the libtool 1.5 distribution (the last one is a copy of
2943 a file called libtool.m4). This seems to have fixed the need to patch
2944 "configure" to support Darwin 1.3 (which I used to do). However, I still
2945 had to patch ltmain.sh to ensure that ${SED} is set (it isn't on my
2946 workstation).
2947
2948 24. Changed the PCRE licence to be the more standard "BSD" licence.
2949
2950
2951 Version 4.5 01-Dec-03
2952 ---------------------
2953
2954 1. There has been some re-arrangement of the code for the match() function so
2955 that it can be compiled in a version that does not call itself recursively.
2956 Instead, it keeps those local variables that need separate instances for
2957 each "recursion" in a frame on the heap, and gets/frees frames whenever it
2958 needs to "recurse". Keeping track of where control must go is done by means
2959 of setjmp/longjmp. The whole thing is implemented by a set of macros that
2960 hide most of the details from the main code, and operates only if
2961 NO_RECURSE is defined while compiling pcre.c. If PCRE is built using the
2962 "configure" mechanism, "--disable-stack-for-recursion" turns on this way of
2963 operating.
2964
2965 To make it easier for callers to provide specially tailored get/free
2966 functions for this usage, two new functions, pcre_stack_malloc, and
2967 pcre_stack_free, are used. They are always called in strict stacking order,
2968 and the size of block requested is always the same.
2969
2970 The PCRE_CONFIG_STACKRECURSE info parameter can be used to find out whether
2971 PCRE has been compiled to use the stack or the heap for recursion. The
2972 -C option of pcretest uses this to show which version is compiled.
2973
2974 A new data escape \S, is added to pcretest; it causes the amounts of store
2975 obtained and freed by both kinds of malloc/free at match time to be added
2976 to the output.
2977
2978 2. Changed the locale test to use "fr_FR" instead of "fr" because that's
2979 what's available on my current Linux desktop machine.
2980
2981 3. When matching a UTF-8 string, the test for a valid string at the start has
2982 been extended. If start_offset is not zero, PCRE now checks that it points
2983 to a byte that is the start of a UTF-8 character. If not, it returns
2984 PCRE_ERROR_BADUTF8_OFFSET (-11). Note: the whole string is still checked;
2985 this is necessary because there may be backward assertions in the pattern.
2986 When matching the same subject several times, it may save resources to use
2987 PCRE_NO_UTF8_CHECK on all but the first call if the string is long.
2988
2989 4. The code for checking the validity of UTF-8 strings has been tightened so
2990 that it rejects (a) strings containing 0xfe or 0xff bytes and (b) strings
2991 containing "overlong sequences".
2992
2993 5. Fixed a bug (appearing twice) that I could not find any way of exploiting!
2994 I had written "if ((digitab[*p++] && chtab_digit) == 0)" where the "&&"
2995 should have been "&", but it just so happened that all the cases this let
2996 through by mistake were picked up later in the function.
2997
2998 6. I had used a variable called "isblank" - this is a C99 function, causing
2999 some compilers to warn. To avoid this, I renamed it (as "blankclass").
3000
3001 7. Cosmetic: (a) only output another newline at the end of pcretest if it is
3002 prompting; (b) run "./pcretest /dev/null" at the start of the test script
3003 so the version is shown; (c) stop "make test" echoing "./RunTest".
3004
3005 8. Added patches from David Burgess to enable PCRE to run on EBCDIC systems.
3006
3007 9. The prototype for memmove() for systems that don't have it was using
3008 size_t, but the inclusion of the header that defines size_t was later. I've
3009 moved the #includes for the C headers earlier to avoid this.
3010
3011 10. Added some adjustments to the code to make it easier to compiler on certain
3012 special systems:
3013
3014 (a) Some "const" qualifiers were missing.
3015 (b) Added the macro EXPORT before all exported functions; by default this
3016 is defined to be empty.
3017 (c) Changed the dftables auxiliary program (that builds chartables.c) so
3018 that it reads its output file name as an argument instead of writing
3019 to the standard output and assuming this can be redirected.
3020
3021 11. In UTF-8 mode, if a recursive reference (e.g. (?1)) followed a character
3022 class containing characters with values greater than 255, PCRE compilation
3023 went into a loop.
3024
3025 12. A recursive reference to a subpattern that was within another subpattern
3026 that had a minimum quantifier of zero caused PCRE to crash. For example,
3027 (x(y(?2))z)? provoked this bug with a subject that got as far as the
3028 recursion. If the recursively-called subpattern itself had a zero repeat,
3029 that was OK.
3030
3031 13. In pcretest, the buffer for reading a data line was set at 30K, but the
3032 buffer into which it was copied (for escape processing) was still set at
3033 1024, so long lines caused crashes.
3034
3035 14. A pattern such as /[ab]{1,3}+/ failed to compile, giving the error
3036 "internal error: code overflow...". This applied to any character class
3037 that was followed by a possessive quantifier.
3038
3039 15. Modified the Makefile to add libpcre.la as a prerequisite for
3040 libpcreposix.la because I was told this is needed for a parallel build to
3041 work.
3042
3043 16. If a pattern that contained .* following optional items at the start was
3044 studied, the wrong optimizing data was generated, leading to matching
3045 errors. For example, studying /[ab]*.*c/ concluded, erroneously, that any
3046 matching string must start with a or b or c. The correct conclusion for
3047 this pattern is that a match can start with any character.
3048
3049
3050 Version 4.4 13-Aug-03
3051 ---------------------
3052
3053 1. In UTF-8 mode, a character class containing characters with values between
3054 127 and 255 was not handled correctly if the compiled pattern was studied.
3055 In fixing this, I have also improved the studying algorithm for such
3056 classes (slightly).
3057
3058 2. Three internal functions had redundant arguments passed to them. Removal
3059 might give a very teeny performance improvement.
3060
3061 3. Documentation bug: the value of the capture_top field in a callout is *one
3062 more than* the number of the hightest numbered captured substring.
3063
3064 4. The Makefile linked pcretest and pcregrep with -lpcre, which could result
3065 in incorrectly linking with a previously installed version. They now link
3066 explicitly with libpcre.la.
3067
3068 5. configure.in no longer needs to recognize Cygwin specially.
3069
3070 6. A problem in pcre.in for Windows platforms is fixed.
3071
3072 7. If a pattern was successfully studied, and the -d (or /D) flag was given to
3073 pcretest, it used to include the size of the study block as part of its
3074 output. Unfortunately, the structure contains a field that has a different
3075 size on different hardware architectures. This meant that the tests that
3076 showed this size failed. As the block is currently always of a fixed size,
3077 this information isn't actually particularly useful in pcretest output, so
3078 I have just removed it.
3079
3080 8. Three pre-processor statements accidentally did not start in column 1.
3081 Sadly, there are *still* compilers around that complain, even though
3082 standard C has not required this for well over a decade. Sigh.
3083
3084 9. In pcretest, the code for checking callouts passed small integers in the
3085 callout_data field, which is a void * field. However, some picky compilers
3086 complained about the casts involved for this on 64-bit systems. Now
3087 pcretest passes the address of the small integer instead, which should get
3088 rid of the warnings.
3089
3090 10. By default, when in UTF-8 mode, PCRE now checks for valid UTF-8 strings at
3091 both compile and run time, and gives an error if an invalid UTF-8 sequence
3092 is found. There is a option for disabling this check in cases where the
3093 string is known to be correct and/or the maximum performance is wanted.
3094
3095 11. In response to a bug report, I changed one line in Makefile.in from
3096
3097 -Wl,--out-implib,.libs/lib@WIN_PREFIX@pcreposix.dll.a \
3098 to
3099 -Wl,--out-implib,.libs/@WIN_PREFIX@libpcreposix.dll.a \
3100
3101 to look similar to other lines, but I have no way of telling whether this
3102 is the right thing to do, as I do not use Windows. No doubt I'll get told
3103 if it's wrong...
3104
3105
3106 Version 4.3 21-May-03
3107 ---------------------
3108
3109 1. Two instances of @WIN_PREFIX@ omitted from the Windows targets in the
3110 Makefile.
3111
3112 2. Some refactoring to improve the quality of the code:
3113
3114 (i) The utf8_table... variables are now declared "const".
3115
3116 (ii) The code for \cx, which used the "case flipping" table to upper case
3117 lower case letters, now just substracts 32. This is ASCII-specific,
3118 but the whole concept of \cx is ASCII-specific, so it seems
3119 reasonable.
3120
3121 (iii) PCRE was using its character types table to recognize decimal and
3122 hexadecimal digits in the pattern. This is silly, because it handles
3123 only 0-9, a-f, and A-F, but the character types table is locale-
3124 specific, which means strange things might happen. A private
3125 table is now used for this - though it costs 256 bytes, a table is
3126 much faster than multiple explicit tests. Of course, the standard
3127 character types table is still used for matching digits in subject
3128 strings against \d.
3129
3130 (iv) Strictly, the identifier ESC_t is reserved by POSIX (all identifiers
3131 ending in _t are). So I've renamed it as ESC_tee.
3132
3133 3. The first argument for regexec() in the POSIX wrapper should have been
3134 defined as "const".
3135
3136 4. Changed pcretest to use malloc() for its buffers so that they can be
3137 Electric Fenced for debugging.
3138
3139 5. There were several places in the code where, in UTF-8 mode, PCRE would try
3140 to read one or more bytes before the start of the subject string. Often this
3141 had no effect on PCRE's behaviour, but in some circumstances it could
3142 provoke a segmentation fault.
3143
3144 6. A lookbehind at the start of a pattern in UTF-8 mode could also cause PCRE
3145 to try to read one or more bytes before the start of the subject string.
3146
3147 7. A lookbehind in a pattern matched in non-UTF-8 mode on a PCRE compiled with
3148 UTF-8 support could misbehave in various ways if the subject string
3149 contained bytes with the 0x80 bit set and the 0x40 bit unset in a lookbehind
3150 area. (PCRE was not checking for the UTF-8 mode flag, and trying to move
3151 back over UTF-8 characters.)
3152
3153
3154 Version 4.2 14-Apr-03
3155 ---------------------
3156
3157 1. Typo "#if SUPPORT_UTF8" instead of "#ifdef SUPPORT_UTF8" fixed.
3158
3159 2. Changes to the building process, supplied by Ronald Landheer-Cieslak
3160 [ON_WINDOWS]: new variable, "#" on non-Windows platforms
3161 [NOT_ON_WINDOWS]: new variable, "#" on Windows platforms
3162 [WIN_PREFIX]: new variable, "cyg" for Cygwin
3163 * Makefile.in: use autoconf substitution for OBJEXT, EXEEXT, BUILD_OBJEXT
3164 and BUILD_EXEEXT
3165 Note: automatic setting of the BUILD variables is not yet working
3166 set CPPFLAGS and BUILD_CPPFLAGS (but don't use yet) - should be used at
3167 compile-time but not at link-time
3168 [LINK]: use for linking executables only
3169 make different versions for Windows and non-Windows
3170 [LINKLIB]: new variable, copy of UNIX-style LINK, used for linking
3171 libraries
3172 [LINK_FOR_BUILD]: new variable
3173 [OBJEXT]: use throughout
3174 [EXEEXT]: use throughout
3175 <winshared>: new target
3176 <wininstall>: new target
3177 <dftables.o>: use native compiler
3178 <dftables>: use native linker
3179 <install>: handle Windows platform correctly
3180 <clean>: ditto
3181 <check>: ditto
3182 copy DLL to top builddir before testing
3183
3184 As part of these changes, -no-undefined was removed again. This was reported
3185 to give trouble on HP-UX 11.0, so getting rid of it seems like a good idea
3186 in any case.
3187
3188 3. Some tidies to get rid of compiler warnings:
3189
3190 . In the match_data structure, match_limit was an unsigned long int, whereas
3191 match_call_count was an int. I've made them both unsigned long ints.
3192
3193 . In pcretest the fact that a const uschar * doesn't automatically cast to
3194 a void * provoked a warning.
3195
3196 . Turning on some more compiler warnings threw up some "shadow" variables
3197 and a few more missing casts.
3198
3199 4. If PCRE was complied with UTF-8 support, but called without the PCRE_UTF8
3200 option, a class that contained a single character with a value between 128
3201 and 255 (e.g. /[\xFF]/) caused PCRE to crash.
3202
3203 5. If PCRE was compiled with UTF-8 support, but called without the PCRE_UTF8
3204 option, a class that contained several characters, but with at least one
3205 whose value was between 128 and 255 caused PCRE to crash.
3206
3207
3208 Version 4.1 12-Mar-03
3209 ---------------------
3210
3211 1. Compiling with gcc -pedantic found a couple of places where casts were
3212 needed, and a string in dftables.c that was longer than standard compilers are
3213 required to support.
3214
3215 2. Compiling with Sun's compiler found a few more places where the code could
3216 be tidied up in order to avoid warnings.
3217
3218 3. The variables for cross-compiling were called HOST_CC and HOST_CFLAGS; the
3219 first of these names is deprecated in the latest Autoconf in favour of the name
3220 CC_FOR_BUILD, because "host" is typically used to mean the system on which the
3221 compiled code will be run. I can't find a reference for HOST_CFLAGS, but by
3222 analogy I have changed it to CFLAGS_FOR_BUILD.
3223
3224 4. Added -no-undefined to the linking command in the Makefile, because this is
3225 apparently helpful for Windows. To make it work, also added "-L. -lpcre" to the
3226 linking step for the pcreposix library.
3227
3228 5. PCRE was failing to diagnose the case of two named groups with the same
3229 name.
3230
3231 6. A problem with one of PCRE's optimizations was discovered. PCRE remembers a
3232 literal character that is needed in the subject for a match, and scans along to
3233 ensure that it is present before embarking on the full matching process. This
3234 saves time in cases of nested unlimited repeats that are never going to match.
3235 Problem: the scan can take a lot of time if the subject is very long (e.g.
3236 megabytes), thus penalizing straightforward matches. It is now done only if the
3237 amount of subject to be scanned is less than 1000 bytes.
3238
3239 7. A lesser problem with the same optimization is that it was recording the
3240 first character of an anchored pattern as "needed", thus provoking a search
3241 right along the subject, even when the first match of the pattern was going to
3242 fail. The "needed" character is now not set for anchored patterns, unless it
3243 follows something in the pattern that is of non-fixed length. Thus, it still
3244 fulfils its original purpose of finding quick non-matches in cases of nested
3245 unlimited repeats, but isn't used for simple anchored patterns such as /^abc/.
3246
3247
3248 Version 4.0 17-Feb-03
3249 ---------------------
3250
3251 1. If a comment in an extended regex that started immediately after a meta-item
3252 extended to the end of string, PCRE compiled incorrect data. This could lead to
3253 all kinds of weird effects. Example: /#/ was bad; /()#/ was bad; /a#/ was not.
3254
3255 2. Moved to autoconf 2.53 and libtool 1.4.2.
3256
3257 3. Perl 5.8 no longer needs "use utf8" for doing UTF-8 things. Consequently,
3258 the special perltest8 script is no longer needed - all the tests can be run
3259 from a single perltest script.
3260
3261 4. From 5.004, Perl has not included the VT character (0x0b) in the set defined
3262 by \s. It has now been removed in PCRE. This means it isn't recognized as
3263 whitespace in /x regexes too, which is the same as Perl. Note that the POSIX
3264 class [:space:] *does* include VT, thereby creating a mess.
3265
3266 5. Added the class [:blank:] (a GNU extension from Perl 5.8) to match only
3267 space and tab.
3268
3269 6. Perl 5.005 was a long time ago. It's time to amalgamate the tests that use
3270 its new features into the main test script, reducing the number of scripts.
3271
3272 7. Perl 5.8 has changed the meaning of patterns like /a(?i)b/. Earlier versions
3273 were backward compatible, and made the (?i) apply to the whole pattern, as if
3274 /i were given. Now it behaves more logically, and applies the option setting
3275 only to what follows. PCRE has been changed to follow suit. However, if it
3276 finds options settings right at the start of the pattern, it extracts them into
3277 the global options, as before. Thus, they show up in the info data.
3278
3279 8. Added support for the \Q...\E escape sequence. Characters in between are
3280 treated as literals. This is slightly different from Perl in that $ and @ are
3281 also handled as literals inside the quotes. In Perl, they will cause variable
3282 interpolation. Note the following examples:
3283
3284 Pattern PCRE matches Perl matches
3285
3286 \Qabc$xyz\E abc$xyz abc followed by the contents of $xyz
3287 \Qabc\$xyz\E abc\$xyz abc\$xyz
3288 \Qabc\E\$\Qxyz\E abc$xyz abc$xyz
3289
3290 For compatibility with Perl, \Q...\E sequences are recognized inside character
3291 classes as well as outside them.
3292
3293 9. Re-organized 3 code statements in pcretest to avoid "overflow in
3294 floating-point constant arithmetic" warnings from a Microsoft compiler. Added a
3295 (size_t) cast to one statement in pcretest and one in pcreposix to avoid
3296 signed/unsigned warnings.
3297
3298 10. SunOS4 doesn't have strtoul(). This was used only for unpicking the -o
3299 option for pcretest, so I've replaced it by a simple function that does just
3300 that job.
3301
3302 11. pcregrep was ending with code 0 instead of 2 for the commands "pcregrep" or
3303 "pcregrep -".
3304
3305 12. Added "possessive quantifiers" ?+, *+, ++, and {,}+ which come from Sun's
3306 Java package. This provides some syntactic sugar for simple cases of what my
3307 documentation calls "once-only subpatterns". A pattern such as x*+ is the same
3308 as (?>x*). In other words, if what is inside (?>...) is just a single repeated
3309 item, you can use this simplified notation. Note that only makes sense with
3310 greedy quantifiers. Consequently, the use of the possessive quantifier forces
3311 greediness, whatever the setting of the PCRE_UNGREEDY option.
3312
3313 13. A change of greediness default within a pattern was not taking effect at
3314 the current level for patterns like /(b+(?U)a+)/. It did apply to parenthesized
3315 subpatterns that followed. Patterns like /b+(?U)a+/ worked because the option
3316 was abstracted outside.
3317
3318 14. PCRE now supports the \G assertion. It is true when the current matching
3319 position is at the start point of the match. This differs from \A when the
3320 starting offset is non-zero. Used with the /g option of pcretest (or similar
3321 code), it works in the same way as it does for Perl's /g option. If all
3322 alternatives of a regex begin with \G, the expression is anchored to the start
3323 match position, and the "anchored" flag is set in the compiled expression.
3324
3325 15. Some bugs concerning the handling of certain option changes within patterns
3326 have been fixed. These applied to options other than (?ims). For example,
3327 "a(?x: b c )d" did not match "XabcdY" but did match "Xa b c dY". It should have
3328 been the other way round. Some of this was related to change 7 above.
3329
3330 16. PCRE now gives errors for /[.x.]/ and /[=x=]/ as unsupported POSIX
3331 features, as Perl does. Previously, PCRE gave the warnings only for /[[.x.]]/
3332 and /[[=x=]]/. PCRE now also gives an error for /[:name:]/ because it supports
3333 POSIX classes only within a class (e.g. /[[:alpha:]]/).
3334
3335 17. Added support for Perl's \C escape. This matches one byte, even in UTF8
3336 mode. Unlike ".", it always matches newline, whatever the setting of
3337 PCRE_DOTALL. However, PCRE does not permit \C to appear in lookbehind
3338 assertions. Perl allows it, but it doesn't (in general) work because it can't
3339 calculate the length of the lookbehind. At least, that's the case for Perl
3340 5.8.0 - I've been told they are going to document that it doesn't work in
3341 future.
3342
3343 18. Added an error diagnosis for escapes that PCRE does not support: these are
3344 \L, \l, \N, \P, \p, \U, \u, and \X.
3345
3346 19. Although correctly diagnosing a missing ']' in a character class, PCRE was
3347 reading past the end of the pattern in cases such as /[abcd/.
3348
3349 20. PCRE was getting more memory than necessary for patterns with classes that
3350 contained both POSIX named classes and other characters, e.g. /[[:space:]abc/.
3351
3352 21. Added some code, conditional on #ifdef VPCOMPAT, to make life easier for
3353 compiling PCRE for use with Virtual Pascal.
3354
3355 22. Small fix to the Makefile to make it work properly if the build is done
3356 outside the source tree.
3357
3358 23. Added a new extension: a condition to go with recursion. If a conditional
3359 subpattern starts with (?(R) the "true" branch is used if recursion has
3360 happened, whereas the "false" branch is used only at the top level.
3361
3362 24. When there was a very long string of literal characters (over 255 bytes
3363 without UTF support, over 250 bytes with UTF support), the computation of how
3364 much memory was required could be incorrect, leading to segfaults or other
3365 strange effects.
3366
3367 25. PCRE was incorrectly assuming anchoring (either to start of subject or to
3368 start of line for a non-DOTALL pattern) when a pattern started with (.*) and
3369 there was a subsequent back reference to those brackets. This meant that, for
3370 example, /(.*)\d+\1/ failed to match "abc123bc". Unfortunately, it isn't
3371 possible to check for precisely this case. All we can do is abandon the
3372 optimization if .* occurs inside capturing brackets when there are any back
3373 references whatsoever. (See below for a better fix that came later.)
3374
3375 26. The handling of the optimization for finding the first character of a
3376 non-anchored pattern, and for finding a character that is required later in the
3377 match were failing in some cases. This didn't break the matching; it just
3378 failed to optimize when it could. The way this is done has been re-implemented.
3379
3380 27. Fixed typo in error message for invalid (?R item (it said "(?p").
3381
3382 28. Added a new feature that provides some of the functionality that Perl
3383 provides with (?{...}). The facility is termed a "callout". The way it is done
3384 in PCRE is for the caller to provide an optional function, by setting
3385 pcre_callout to its entry point. Like pcre_malloc and pcre_free, this is a
3386 global variable. By default it is unset, which disables all calling out. To get
3387 the function called, the regex must include (?C) at appropriate points. This
3388 is, in fact, equivalent to (?C0), and any number <= 255 may be given with (?C).
3389 This provides a means of identifying different callout points. When PCRE
3390 reaches such a point in the regex, if pcre_callout has been set, the external
3391 function is called. It is provided with data in a structure called
3392 pcre_callout_block, which is defined in pcre.h. If the function returns 0,
3393 matching continues; if it returns a non-zero value, the match at the current
3394 point fails. However, backtracking will occur if possible. [This was changed
3395 later and other features added - see item 49 below.]
3396
3397 29. pcretest is upgraded to test the callout functionality. It provides a
3398 callout function that displays information. By default, it shows the start of
3399 the match and the current position in the text. There are some new data escapes
3400 to vary what happens:
3401
3402 \C+ in addition, show current contents of captured substrings
3403 \C- do not supply a callout function
3404 \C!n return 1 when callout number n is reached
3405 \C!n!m return 1 when callout number n is reached for the mth time
3406
3407 30. If pcregrep was called with the -l option and just a single file name, it
3408 output "<stdin>" if a match was found, instead of the file name.
3409
3410 31. Improve the efficiency of the POSIX API to PCRE. If the number of capturing
3411 slots is less than POSIX_MALLOC_THRESHOLD, use a block on the stack to pass to
3412 pcre_exec(). This saves a malloc/free per call. The default value of
3413 POSIX_MALLOC_THRESHOLD is 10; it can be changed by --with-posix-malloc-threshold
3414 when configuring.
3415
3416 32. The default maximum size of a compiled pattern is 64K. There have been a
3417 few cases of people hitting this limit. The code now uses macros to handle the
3418 storing of links as offsets within the compiled pattern. It defaults to 2-byte
3419 links, but this can be changed to 3 or 4 bytes by --with-link-size when
3420 configuring. Tests 2 and 5 work only with 2-byte links because they output
3421 debugging information about compiled patterns.
3422
3423 33. Internal code re-arrangements:
3424
3425 (a) Moved the debugging function for printing out a compiled regex into
3426 its own source file (printint.c) and used #include to pull it into
3427 pcretest.c and, when DEBUG is defined, into pcre.c, instead of having two
3428 separate copies.
3429
3430 (b) Defined the list of op-code names for debugging as a macro in
3431 internal.h so that it is next to the definition of the opcodes.
3432
3433 (c) Defined a table of op-code lengths for simpler skipping along compiled
3434 code. This is again a macro in internal.h so that it is next to the
3435 definition of the opcodes.
3436
3437 34. Added support for recursive calls to individual subpatterns, along the
3438 lines of Robin Houston's patch (but implemented somewhat differently).
3439
3440 35. Further mods to the Makefile to help Win32. Also, added code to pcregrep to
3441 allow it to read and process whole directories in Win32. This code was
3442 contributed by Lionel Fourquaux; it has not been tested by me.
3443
3444 36. Added support for named subpatterns. The Python syntax (?P<name>...) is
3445 used to name a group. Names consist of alphanumerics and underscores, and must
3446 be unique. Back references use the syntax (?P=name) and recursive calls use
3447 (?P>name) which is a PCRE extension to the Python extension. Groups still have
3448 numbers. The function pcre_fullinfo() can be used after compilation to extract
3449 a name/number map. There are three relevant calls:
3450
3451 PCRE_INFO_NAMEENTRYSIZE yields the size of each entry in the map
3452 PCRE_INFO_NAMECOUNT yields the number of entries
3453 PCRE_INFO_NAMETABLE yields a pointer to the map.
3454
3455 The map is a vector of fixed-size entries. The size of each entry depends on
3456 the length of the longest name used. The first two bytes of each entry are the
3457 group number, most significant byte first. There follows the corresponding
3458 name, zero terminated. The names are in alphabetical order.
3459
3460 37. Make the maximum literal string in the compiled code 250 for the non-UTF-8
3461 case instead of 255. Making it the same both with and without UTF-8 support
3462 means that the same test output works with both.
3463
3464 38. There was a case of malloc(0) in the POSIX testing code in pcretest. Avoid
3465 calling malloc() with a zero argument.
3466
3467 39. Change 25 above had to resort to a heavy-handed test for the .* anchoring
3468 optimization. I've improved things by keeping a bitmap of backreferences with
3469 numbers 1-31 so that if .* occurs inside capturing brackets that are not in
3470 fact referenced, the optimization can be applied. It is unlikely that a
3471 relevant occurrence of .* (i.e. one which might indicate anchoring or forcing
3472 the match to follow \n) will appear inside brackets with a number greater than
3473 31, but if it does, any back reference > 31 suppresses the optimization.
3474
3475 40. Added a new compile-time option PCRE_NO_AUTO_CAPTURE. This has the effect
3476 of disabling numbered capturing parentheses. Any opening parenthesis that is
3477 not followed by ? behaves as if it were followed by ?: but named parentheses
3478 can still be used for capturing (and they will acquire numbers in the usual
3479 way).
3480
3481 41. Redesigned the return codes from the match() function into yes/no/error so
3482 that errors can be passed back from deep inside the nested calls. A malloc
3483 failure while inside a recursive subpattern call now causes the
3484 PCRE_ERROR_NOMEMORY return instead of quietly going wrong.
3485
3486 42. It is now possible to set a limit on the number of times the match()
3487 function is called in a call to pcre_exec(). This facility makes it possible to
3488 limit the amount of recursion and backtracking, though not in a directly
3489 obvious way, because the match() function is used in a number of different
3490 circumstances. The count starts from zero for each position in the subject
3491 string (for non-anchored patterns). The default limit is, for compatibility, a
3492 large number, namely 10 000 000. You can change this in two ways:
3493
3494 (a) When configuring PCRE before making, you can use --with-match-limit=n
3495 to set a default value for the compiled library.
3496
3497 (b) For each call to pcre_exec(), you can pass a pcre_extra block in which
3498 a different value is set. See 45 below.
3499
3500 If the limit is exceeded, pcre_exec() returns PCRE_ERROR_MATCHLIMIT.
3501
3502 43. Added a new function pcre_config(int, void *) to enable run-time extraction
3503 of things that can be changed at compile time. The first argument specifies
3504 what is wanted and the second points to where the information is to be placed.
3505 The current list of available information is:
3506
3507 PCRE_CONFIG_UTF8
3508
3509 The output is an integer that is set to one if UTF-8 support is available;
3510 otherwise it is set to zero.
3511
3512 PCRE_CONFIG_NEWLINE
3513
3514 The output is an integer that it set to the value of the code that is used for
3515 newline. It is either LF (10) or CR (13).
3516
3517 PCRE_CONFIG_LINK_SIZE
3518
3519 The output is an integer that contains the number of bytes used for internal
3520 linkage in compiled expressions. The value is 2, 3, or 4. See item 32 above.
3521
3522 PCRE_CONFIG_POSIX_MALLOC_THRESHOLD
3523
3524 The output is an integer that contains the threshold above which the POSIX
3525 interface uses malloc() for output vectors. See item 31 above.
3526
3527 PCRE_CONFIG_MATCH_LIMIT
3528
3529 The output is an unsigned integer that contains the default limit of the number
3530 of match() calls in a pcre_exec() execution. See 42 above.
3531
3532 44. pcretest has been upgraded by the addition of the -C option. This causes it
3533 to extract all the available output from the new pcre_config() function, and to
3534 output it. The program then exits immediately.
3535
3536 45. A need has arisen to pass over additional data with calls to pcre_exec() in
3537 order to support additional features. One way would have been to define
3538 pcre_exec2() (for example) with extra arguments, but this would not have been
3539 extensible, and would also have required all calls to the original function to
3540 be mapped to the new one. Instead, I have chosen to extend the mechanism that
3541 is used for passing in "extra" data from pcre_study().
3542
3543 The pcre_extra structure is now exposed and defined in pcre.h. It currently
3544 contains the following fields:
3545
3546 flags a bitmap indicating which of the following fields are set
3547 study_data opaque data from pcre_study()
3548 match_limit a way of specifying a limit on match() calls for a specific
3549 call to pcre_exec()
3550 callout_data data for callouts (see 49 below)
3551
3552 The flag bits are also defined in pcre.h, and are
3553
3554 PCRE_EXTRA_STUDY_DATA
3555 PCRE_EXTRA_MATCH_LIMIT
3556 PCRE_EXTRA_CALLOUT_DATA
3557
3558 The pcre_study() function now returns one of these new pcre_extra blocks, with
3559 the actual study data pointed to by the study_data field, and the
3560 PCRE_EXTRA_STUDY_DATA flag set. This can be passed directly to pcre_exec() as
3561 before. That is, this change is entirely upwards-compatible and requires no
3562 change to existing code.
3563
3564 If you want to pass in additional data to pcre_exec(), you can either place it
3565 in a pcre_extra block provided by pcre_study(), or create your own pcre_extra
3566 block.
3567
3568 46. pcretest has been extended to test the PCRE_EXTRA_MATCH_LIMIT feature. If a
3569 data string contains the escape sequence \M, pcretest calls pcre_exec() several
3570 times with different match limits, until it finds the minimum value needed for
3571 pcre_exec() to complete. The value is then output. This can be instructive; for
3572 most simple matches the number is quite small, but for pathological cases it
3573 gets very large very quickly.
3574
3575 47. There's a new option for pcre_fullinfo() called PCRE_INFO_STUDYSIZE. It
3576 returns the size of the data block pointed to by the study_data field in a
3577 pcre_extra block, that is, the value that was passed as the argument to
3578 pcre_malloc() when PCRE was getting memory in which to place the information
3579 created by pcre_study(). The fourth argument should point to a size_t variable.
3580 pcretest has been extended so that this information is shown after a successful
3581 pcre_study() call when information about the compiled regex is being displayed.
3582
3583 48. Cosmetic change to Makefile: there's no need to have / after $(DESTDIR)
3584 because what follows is always an absolute path. (Later: it turns out that this
3585 is more than cosmetic for MinGW, because it doesn't like empty path
3586 components.)
3587
3588 49. Some changes have been made to the callout feature (see 28 above):
3589
3590 (i) A callout function now has three choices for what it returns:
3591
3592 0 => success, carry on matching
3593 > 0 => failure at this point, but backtrack if possible
3594 < 0 => serious error, return this value from pcre_exec()
3595
3596 Negative values should normally be chosen from the set of PCRE_ERROR_xxx
3597 values. In particular, returning PCRE_ERROR_NOMATCH forces a standard
3598 "match failed" error. The error number PCRE_ERROR_CALLOUT is reserved for
3599 use by callout functions. It will never be used by PCRE itself.
3600
3601 (ii) The pcre_extra structure (see 45 above) has a void * field called
3602 callout_data, with corresponding flag bit PCRE_EXTRA_CALLOUT_DATA. The
3603 pcre_callout_block structure has a field of the same name. The contents of
3604 the field passed in the pcre_extra structure are passed to the callout
3605 function in the corresponding field in the callout block. This makes it
3606 easier to use the same callout-containing regex from multiple threads. For
3607 testing, the pcretest program has a new data escape
3608
3609 \C*n pass the number n (may be negative) as callout_data
3610
3611 If the callout function in pcretest receives a non-zero value as
3612 callout_data, it returns that value.
3613
3614 50. Makefile wasn't handling CFLAGS properly when compiling dftables. Also,
3615 there were some redundant $(CFLAGS) in commands that are now specified as
3616 $(LINK), which already includes $(CFLAGS).
3617
3618 51. Extensions to UTF-8 support are listed below. These all apply when (a) PCRE
3619 has been compiled with UTF-8 support *and* pcre_compile() has been compiled
3620 with the PCRE_UTF8 flag. Patterns that are compiled without that flag assume
3621 one-byte characters throughout. Note that case-insensitive matching applies
3622 only to characters whose values are less than 256. PCRE doesn't support the
3623 notion of cases for higher-valued characters.
3624
3625 (i) A character class whose characters are all within 0-255 is handled as
3626 a bit map, and the map is inverted for negative classes. Previously, a
3627 character > 255 always failed to match such a class; however it should
3628 match if the class was a negative one (e.g. [^ab]). This has been fixed.
3629
3630 (ii) A negated character class with a single character < 255 is coded as
3631 "not this character" (OP_NOT). This wasn't working properly when the test
3632 character was multibyte, either singly or repeated.
3633
3634 (iii) Repeats of multibyte characters are now handled correctly in UTF-8
3635 mode, for example: \x{100}{2,3}.
3636
3637 (iv) The character escapes \b, \B, \d, \D, \s, \S, \w, and \W (either
3638 singly or repeated) now correctly test multibyte characters. However,
3639 PCRE doesn't recognize any characters with values greater than 255 as
3640 digits, spaces, or word characters. Such characters always match \D, \S,
3641 and \W, and never match \d, \s, or \w.
3642
3643 (v) Classes may now contain characters and character ranges with values
3644 greater than 255. For example: [ab\x{100}-\x{400}].
3645
3646 (vi) pcregrep now has a --utf-8 option (synonym -u) which makes it call
3647 PCRE in UTF-8 mode.
3648
3649 52. The info request value PCRE_INFO_FIRSTCHAR has been renamed
3650 PCRE_INFO_FIRSTBYTE because it is a byte value. However, the old name is
3651 retained for backwards compatibility. (Note that LASTLITERAL is also a byte
3652 value.)
3653
3654 53. The single man page has become too large. I have therefore split it up into
3655 a number of separate man pages. These also give rise to individual HTML pages;
3656 these are now put in a separate directory, and there is an index.html page that
3657 lists them all. Some hyperlinking between the pages has been installed.
3658
3659 54. Added convenience functions for handling named capturing parentheses.
3660
3661 55. Unknown escapes inside character classes (e.g. [\M]) and escapes that
3662 aren't interpreted therein (e.g. [\C]) are literals in Perl. This is now also
3663 true in PCRE, except when the PCRE_EXTENDED option is set, in which case they
3664 are faulted.
3665
3666 56. Introduced HOST_CC and HOST_CFLAGS which can be set in the environment when
3667 calling configure. These values are used when compiling the dftables.c program
3668 which is run to generate the source of the default character tables. They
3669 default to the values of CC and CFLAGS. If you are cross-compiling PCRE,
3670 you will need to set these values.
3671
3672 57. Updated the building process for Windows DLL, as provided by Fred Cox.
3673
3674
3675 Version 3.9 02-Jan-02
3676 ---------------------
3677
3678 1. A bit of extraneous text had somehow crept into the pcregrep documentation.
3679
3680 2. If --disable-static was given, the building process failed when trying to
3681 build pcretest and pcregrep. (For some reason it was using libtool to compile
3682 them, which is not right, as they aren't part of the library.)
3683
3684
3685 Version 3.8 18-Dec-01
3686 ---------------------
3687
3688 1. The experimental UTF-8 code was completely screwed up. It was packing the
3689 bytes in the wrong order. How dumb can you get?
3690
3691
3692 Version 3.7 29-Oct-01
3693 ---------------------
3694
3695 1. In updating pcretest to check change 1 of version 3.6, I screwed up.
3696 This caused pcretest, when used on the test data, to segfault. Unfortunately,
3697 this didn't happen under Solaris 8, where I normally test things.
3698
3699 2. The Makefile had to be changed to make it work on BSD systems, where 'make'
3700 doesn't seem to recognize that ./xxx and xxx are the same file. (This entry
3701 isn't in ChangeLog distributed with 3.7 because I forgot when I hastily made
3702 this fix an hour or so after the initial 3.7 release.)
3703
3704
3705 Version 3.6 23-Oct-01
3706 ---------------------
3707
3708 1. Crashed with /(sens|respons)e and \1ibility/ and "sense and sensibility" if
3709 offsets passed as NULL with zero offset count.
3710
3711 2. The config.guess and config.sub files had not been updated when I moved to
3712 the latest autoconf.
3713
3714
3715 Version 3.5 15-Aug-01
3716 ---------------------
3717
3718 1. Added some missing #if !defined NOPOSIX conditionals in pcretest.c that
3719 had been forgotten.
3720
3721 2. By using declared but undefined structures, we can avoid using "void"
3722 definitions in pcre.h while keeping the internal definitions of the structures
3723 private.
3724
3725 3. The distribution is now built using autoconf 2.50 and libtool 1.4. From a
3726 user point of view, this means that both static and shared libraries are built
3727 by default, but this can be individually controlled. More of the work of
3728 handling this static/shared cases is now inside libtool instead of PCRE's make
3729 file.
3730
3731 4. The pcretest utility is now installed along with pcregrep because it is
3732 useful for users (to test regexs) and by doing this, it automatically gets
3733 relinked by libtool. The documentation has been turned into a man page, so
3734 there are now .1, .txt, and .html versions in /doc.
3735
3736 5. Upgrades to pcregrep:
3737 (i) Added long-form option names like gnu grep.
3738 (ii) Added --help to list all options with an explanatory phrase.
3739 (iii) Added -r, --recursive to recurse into sub-directories.
3740 (iv) Added -f, --file to read patterns from a file.
3741
3742 6. pcre_exec() was referring to its "code" argument before testing that
3743 argument for NULL (and giving an error if it was NULL).
3744
3745 7. Upgraded Makefile.in to allow for compiling in a different directory from
3746 the source directory.
3747
3748 8. Tiny buglet in pcretest: when pcre_fullinfo() was called to retrieve the
3749 options bits, the pointer it was passed was to an int instead of to an unsigned
3750 long int. This mattered only on 64-bit systems.
3751
3752 9. Fixed typo (3.4/1) in pcre.h again. Sigh. I had changed pcre.h (which is
3753 generated) instead of pcre.in, which it its source. Also made the same change
3754 in several of the .c files.
3755
3756 10. A new release of gcc defines printf() as a macro, which broke pcretest
3757 because it had an ifdef in the middle of a string argument for printf(). Fixed
3758 by using separate calls to printf().
3759
3760 11. Added --enable-newline-is-cr and --enable-newline-is-lf to the configure
3761 script, to force use of CR or LF instead of \n in the source. On non-Unix
3762 systems, the value can be set in config.h.
3763
3764 12. The limit of 200 on non-capturing parentheses is a _nesting_ limit, not an
3765 absolute limit. Changed the text of the error message to make this clear, and
3766 likewise updated the man page.
3767
3768 13. The limit of 99 on the number of capturing subpatterns has been removed.
3769 The new limit is 65535, which I hope will not be a "real" limit.
3770
3771
3772 Version 3.4 22-Aug-00
3773 ---------------------
3774
3775 1. Fixed typo in pcre.h: unsigned const char * changed to const unsigned char *.
3776
3777 2. Diagnose condition (?(0) as an error instead of crashing on matching.
3778
3779
3780 Version 3.3 01-Aug-00
3781 ---------------------
3782
3783 1. If an octal character was given, but the value was greater than \377, it
3784 was not getting masked to the least significant bits, as documented. This could
3785 lead to crashes in some systems.
3786
3787 2. Perl 5.6 (if not earlier versions) accepts classes like [a-\d] and treats
3788 the hyphen as a literal. PCRE used to give an error; it now behaves like Perl.
3789
3790 3. Added the functions pcre_free_substring() and pcre_free_substring_list().
3791 These just pass their arguments on to (pcre_free)(), but they are provided
3792 because some uses of PCRE bind it to non-C systems that can call its functions,
3793 but cannot call free() or pcre_free() directly.
3794
3795 4. Add "make test" as a synonym for "make check". Corrected some comments in
3796 the Makefile.
3797
3798 5. Add $(DESTDIR)/ in front of all the paths in the "install" target in the
3799 Makefile.
3800
3801 6. Changed the name of pgrep to pcregrep, because Solaris has introduced a
3802 command called pgrep for grepping around the active processes.
3803
3804 7. Added the beginnings of support for UTF-8 character strings.
3805
3806 8. Arranged for the Makefile to pass over the settings of CC, CFLAGS, and
3807 RANLIB to ./ltconfig so that they are used by libtool. I think these are all
3808 the relevant ones. (AR is not passed because ./ltconfig does its own figuring
3809 out for the ar command.)
3810
3811
3812 Version 3.2 12-May-00
3813 ---------------------
3814
3815 This is purely a bug fixing release.
3816
3817 1. If the pattern /((Z)+|A)*/ was matched agained ZABCDEFG it matched Z instead
3818 of ZA. This was just one example of several cases that could provoke this bug,
3819 which was introduced by change 9 of version 2.00. The code for breaking
3820 infinite loops after an iteration that matches an empty string was't working
3821 correctly.
3822
3823 2. The pcretest program was not imitating Perl correctly for the pattern /a*/g
3824 when matched against abbab (for example). After matching an empty string, it
3825 wasn't forcing anchoring when setting PCRE_NOTEMPTY for the next attempt; this
3826 caused it to match further down the string than it should.
3827
3828 3. The code contained an inclusion of sys/types.h. It isn't clear why this
3829 was there because it doesn't seem to be needed, and it causes trouble on some
3830 systems, as it is not a Standard C header. It has been removed.
3831
3832 4. Made 4 silly changes to the source to avoid stupid compiler warnings that
3833 were reported on the Macintosh. The changes were from
3834
3835 while ((c = *(++ptr)) != 0 && c != '\n');
3836 to
3837 while ((c = *(++ptr)) != 0 && c != '\n') ;
3838
3839 Totally extraordinary, but if that's what it takes...
3840
3841 5. PCRE is being used in one environment where neither memmove() nor bcopy() is
3842 available. Added HAVE_BCOPY and an autoconf test for it; if neither
3843 HAVE_MEMMOVE nor HAVE_BCOPY is set, use a built-in emulation function which
3844 assumes the way PCRE uses memmove() (always moving upwards).
3845
3846 6. PCRE is being used in one environment where strchr() is not available. There
3847 was only one use in pcre.c, and writing it out to avoid strchr() probably gives
3848 faster code anyway.
3849
3850
3851 Version 3.1 09-Feb-00
3852 ---------------------
3853
3854 The only change in this release is the fixing of some bugs in Makefile.in for
3855 the "install" target:
3856
3857 (1) It was failing to install pcreposix.h.
3858
3859 (2) It was overwriting the pcre.3 man page with the pcreposix.3 man page.
3860
3861
3862 Version 3.0 01-Feb-00
3863 ---------------------
3864
3865 1. Add support for the /+ modifier to perltest (to output $` like it does in
3866 pcretest).
3867
3868 2. Add support for the /g modifier to perltest.
3869
3870 3. Fix pcretest so that it behaves even more like Perl for /g when the pattern
3871 matches null strings.
3872
3873 4. Fix perltest so that it doesn't do unwanted things when fed an empty
3874 pattern. Perl treats empty patterns specially - it reuses the most recent
3875 pattern, which is not what we want. Replace // by /(?#)/ in order to avoid this
3876 effect.
3877
3878 5. The POSIX interface was broken in that it was just handing over the POSIX
3879 captured string vector to pcre_exec(), but (since release 2.00) PCRE has
3880 required a bigger vector, with some working space on the end. This means that
3881 the POSIX wrapper now has to get and free some memory, and copy the results.
3882
3883 6. Added some simple autoconf support, placing the test data and the
3884 documentation in separate directories, re-organizing some of the
3885 information files, and making it build pcre-config (a GNU standard). Also added
3886 libtool support for building PCRE as a shared library, which is now the
3887 default.
3888
3889 7. Got rid of the leading zero in the definition of PCRE_MINOR because 08 and
3890 09 are not valid octal constants. Single digits will be used for minor values
3891 less than 10.
3892
3893 8. Defined REG_EXTENDED and REG_NOSUB as zero in the POSIX header, so that
3894 existing programs that set these in the POSIX interface can use PCRE without
3895 modification.
3896
3897 9. Added a new function, pcre_fullinfo() with an extensible interface. It can
3898 return all that pcre_info() returns, plus additional data. The pcre_info()
3899 function is retained for compatibility, but is considered to be obsolete.
3900
3901 10. Added experimental recursion feature (?R) to handle one common case that
3902 Perl 5.6 will be able to do with (?p{...}).
3903
3904 11. Added support for POSIX character classes like [:alpha:], which Perl is
3905 adopting.
3906
3907
3908 Version 2.08 31-Aug-99
3909 ----------------------
3910
3911 1. When startoffset was not zero and the pattern began with ".*", PCRE was not
3912 trying to match at the startoffset position, but instead was moving forward to
3913 the next newline as if a previous match had failed.
3914
3915 2. pcretest was not making use of PCRE_NOTEMPTY when repeating for /g and /G,
3916 and could get into a loop if a null string was matched other than at the start
3917 of the subject.
3918
3919 3. Added definitions of PCRE_MAJOR and PCRE_MINOR to pcre.h so the version can
3920 be distinguished at compile time, and for completeness also added PCRE_DATE.
3921
3922 5. Added Paul Sokolovsky's minor changes to make it easy to compile a Win32 DLL
3923 in GnuWin32 environments.
3924
3925
3926 Version 2.07 29-Jul-99
3927 ----------------------
3928
3929 1. The documentation is now supplied in plain text form and HTML as well as in
3930 the form of man page sources.
3931
3932 2. C++ compilers don't like assigning (void *) values to other pointer types.
3933 In particular this affects malloc(). Although there is no problem in Standard
3934 C, I've put in casts to keep C++ compilers happy.
3935
3936 3. Typo on pcretest.c; a cast of (unsigned char *) in the POSIX regexec() call
3937 should be (const char *).
3938
3939 4. If NOPOSIX is defined, pcretest.c compiles without POSIX support. This may
3940 be useful for non-Unix systems who don't want to bother with the POSIX stuff.
3941 However, I haven't made this a standard facility. The documentation doesn't
3942 mention it, and the Makefile doesn't support it.
3943
3944 5. The Makefile now contains an "install" target, with editable destinations at
3945 the top of the file. The pcretest program is not installed.
3946
3947 6. pgrep -V now gives the PCRE version number and date.
3948
3949 7. Fixed bug: a zero repetition after a literal string (e.g. /abcde{0}/) was
3950 causing the entire string to be ignored, instead of just the last character.
3951
3952 8. If a pattern like /"([^\\"]+|\\.)*"/ is applied in the normal way to a
3953 non-matching string, it can take a very, very long time, even for strings of
3954 quite modest length, because of the nested recursion. PCRE now does better in
3955 some of these cases. It does this by remembering the last required literal
3956 character in the pattern, and pre-searching the subject to ensure it is present
3957 before running the real match. In other words, it applies a heuristic to detect
3958 some types of certain failure quickly, and in the above example, if presented
3959 with a string that has no trailing " it gives "no match" very quickly.
3960
3961 9. A new runtime option PCRE_NOTEMPTY causes null string matches to be ignored;
3962 other alternatives are tried instead.
3963
3964
3965 Version 2.06 09-Jun-99
3966 ----------------------
3967
3968 1. Change pcretest's output for amount of store used to show just the code
3969 space, because the remainder (the data block) varies in size between 32-bit and
3970 64-bit systems.
3971
3972 2. Added an extra argument to pcre_exec() to supply an offset in the subject to
3973 start matching at. This allows lookbehinds to work when searching for multiple
3974 occurrences in a string.
3975
3976 3. Added additional options to pcretest for testing multiple occurrences:
3977
3978 /+ outputs the rest of the string that follows a match
3979 /g loops for multiple occurrences, using the new startoffset argument
3980 /G loops for multiple occurrences by passing an incremented pointer
3981
3982 4. PCRE wasn't doing the "first character" optimization for patterns starting
3983 with \b or \B, though it was doing it for other lookbehind assertions. That is,
3984 it wasn't noticing that a match for a pattern such as /\bxyz/ has to start with
3985 the letter 'x'. On long subject strings, this gives a significant speed-up.
3986
3987
3988 Version 2.05 21-Apr-99
3989 ----------------------
3990
3991 1. Changed the type of magic_number from int to long int so that it works
3992 properly on 16-bit systems.
3993
3994 2. Fixed a bug which caused patterns starting with .* not to work correctly
3995 when the subject string contained newline characters. PCRE was assuming
3996 anchoring for such patterns in all cases, which is not correct because .* will
3997 not pass a newline unless PCRE_DOTALL is set. It now assumes anchoring only if
3998 DOTALL is set at top level; otherwise it knows that patterns starting with .*
3999 must be retried after every newline in the subject.
4000
4001
4002 Version 2.04 18-Feb-99
4003 ----------------------
4004
4005 1. For parenthesized subpatterns with repeats whose minimum was zero, the
4006 computation of the store needed to hold the pattern was incorrect (too large).
4007 If such patterns were nested a few deep, this could multiply and become a real
4008 problem.
4009
4010 2. Added /M option to pcretest to show the memory requirement of a specific
4011 pattern. Made -m a synonym of -s (which does this globally) for compatibility.
4012
4013 3. Subpatterns of the form (regex){n,m} (i.e. limited maximum) were being
4014 compiled in such a way that the backtracking after subsequent failure was
4015 pessimal. Something like (a){0,3} was compiled as (a)?(a)?(a)? instead of
4016 ((a)((a)(a)?)?)? with disastrous performance if the maximum was of any size.
4017
4018
4019 Version 2.03 02-Feb-99
4020 ----------------------
4021
4022 1. Fixed typo and small mistake in man page.
4023
4024 2. Added 4th condition (GPL supersedes if conflict) and created separate
4025 LICENCE file containing the conditions.
4026
4027 3. Updated pcretest so that patterns such as /abc\/def/ work like they do in
4028 Perl, that is the internal \ allows the delimiter to be included in the
4029 pattern. Locked out the use of \ as a delimiter. If \ immediately follows
4030 the final delimiter, add \ to the end of the pattern (to test the error).
4031
4032 4. Added the convenience functions for extracting substrings after a successful
4033 match. Updated pcretest to make it able to test these functions.
4034
4035
4036 Version 2.02 14-Jan-99
4037 ----------------------
4038
4039 1. Initialized the working variables associated with each extraction so that
4040 their saving and restoring doesn't refer to uninitialized store.
4041
4042 2. Put dummy code into study.c in order to trick the optimizer of the IBM C
4043 compiler for OS/2 into generating correct code. Apparently IBM isn't going to
4044 fix the problem.
4045
4046 3. Pcretest: the timing code wasn't using LOOPREPEAT for timing execution
4047 calls, and wasn't printing the correct value for compiling calls. Increased the
4048 default value of LOOPREPEAT, and the number of significant figures in the
4049 times.
4050
4051 4. Changed "/bin/rm" in the Makefile to "-rm" so it works on Windows NT.
4052
4053 5. Renamed "deftables" as "dftables" to get it down to 8 characters, to avoid
4054 a building problem on Windows NT with a FAT file system.
4055
4056
4057 Version 2.01 21-Oct-98
4058 ----------------------
4059
4060 1. Changed the API for pcre_compile() to allow for the provision of a pointer
4061 to character tables built by pcre_maketables() in the current locale. If NULL
4062 is passed, the default tables are used.
4063
4064
4065 Version 2.00 24-Sep-98
4066 ----------------------
4067
4068 1. Since the (>?) facility is in Perl 5.005, don't require PCRE_EXTRA to enable
4069 it any more.
4070
4071 2. Allow quantification of (?>) groups, and make it work correctly.
4072
4073 3. The first character computation wasn't working for (?>) groups.
4074
4075 4. Correct the implementation of \Z (it is permitted to match on the \n at the
4076 end of the subject) and add 5.005's \z, which really does match only at the
4077 very end of the subject.
4078
4079 5. Remove the \X "cut" facility; Perl doesn't have it, and (?> is neater.
4080
4081 6. Remove the ability to specify CASELESS, MULTILINE, DOTALL, and
4082 DOLLAR_END_ONLY at runtime, to make it possible to implement the Perl 5.005
4083 localized options. All options to pcre_study() were also removed.
4084
4085 7. Add other new features from 5.005:
4086
4087 $(?<= positive lookbehind
4088 $(?<! negative lookbehind
4089 (?imsx-imsx) added the unsetting capability
4090 such a setting is global if at outer level; local otherwise
4091 (?imsx-imsx:) non-capturing groups with option setting
4092 (?(cond)re|re) conditional pattern matching
4093
4094 A backreference to itself in a repeated group matches the previous
4095 captured string.
4096
4097 8. General tidying up of studying (both automatic and via "study")
4098 consequential on the addition of new assertions.
4099
4100 9. As in 5.005, unlimited repeated groups that could match an empty substring
4101 are no longer faulted at compile time. Instead, the loop is forcibly broken at
4102 runtime if any iteration does actually match an empty substring.
4103
4104 10. Include the RunTest script in the distribution.
4105
4106 11. Added tests from the Perl 5.005_02 distribution. This showed up a few
4107 discrepancies, some of which were old and were also with respect to 5.004. They
4108 have now been fixed.
4109
4110
4111 Version 1.09 28-Apr-98
4112 ----------------------
4113
4114 1. A negated single character class followed by a quantifier with a minimum
4115 value of one (e.g. [^x]{1,6} ) was not compiled correctly. This could lead to
4116 program crashes, or just wrong answers. This did not apply to negated classes
4117 containing more than one character, or to minima other than one.
4118
4119
4120 Version 1.08 27-Mar-98
4121 ----------------------
4122
4123 1. Add PCRE_UNGREEDY to invert the greediness of quantifiers.
4124
4125 2. Add (?U) and (?X) to set PCRE_UNGREEDY and PCRE_EXTRA respectively. The
4126 latter must appear before anything that relies on it in the pattern.
4127
4128
4129 Version 1.07 16-Feb-98
4130 ----------------------
4131
4132 1. A pattern such as /((a)*)*/ was not being diagnosed as in error (unlimited
4133 repeat of a potentially empty string).
4134
4135
4136 Version 1.06 23-Jan-98
4137 ----------------------
4138
4139 1. Added Markus Oberhumer's little patches for C++.
4140
4141 2. Literal strings longer than 255 characters were broken.
4142
4143
4144 Version 1.05 23-Dec-97
4145 ----------------------
4146
4147 1. Negated character classes containing more than one character were failing if
4148 PCRE_CASELESS was set at run time.
4149
4150
4151 Version 1.04 19-Dec-97
4152 ----------------------
4153
4154 1. Corrected the man page, where some "const" qualifiers had been omitted.
4155
4156 2. Made debugging output print "{0,xxx}" instead of just "{,xxx}" to agree with
4157 input syntax.
4158
4159 3. Fixed memory leak which occurred when a regex with back references was
4160 matched with an offsets vector that wasn't big enough. The temporary memory
4161 that is used in this case wasn't being freed if the match failed.
4162
4163 4. Tidied pcretest to ensure it frees memory that it gets.
4164
4165 5. Temporary memory was being obtained in the case where the passed offsets
4166 vector was exactly big enough.
4167
4168 6. Corrected definition of offsetof() from change 5 below.
4169
4170 7. I had screwed up change 6 below and broken the rules for the use of
4171 setjmp(). Now fixed.
4172
4173
4174 Version 1.03 18-Dec-97
4175 ----------------------
4176
4177 1. A erroneous regex with a missing opening parenthesis was correctly
4178 diagnosed, but PCRE attempted to access brastack[-1], which could cause crashes
4179 on some systems.
4180
4181 2. Replaced offsetof(real_pcre, code) by offsetof(real_pcre, code[0]) because
4182 it was reported that one broken compiler failed on the former because "code" is
4183 also an independent variable.
4184
4185 3. The erroneous regex a[]b caused an array overrun reference.
4186
4187 4. A regex ending with a one-character negative class (e.g. /[^k]$/) did not
4188 fail on data ending with that character. (It was going on too far, and checking
4189 the next character, typically a binary zero.) This was specific to the
4190 optimized code for single-character negative classes.
4191
4192 5. Added a contributed patch from the TIN world which does the following:
4193
4194 + Add an undef for memmove, in case the the system defines a macro for it.
4195
4196 + Add a definition of offsetof(), in case there isn't one. (I don't know
4197 the reason behind this - offsetof() is part of the ANSI standard - but
4198 it does no harm).
4199
4200 + Reduce the ifdef's in pcre.c using macro DPRINTF, thereby eliminating
4201 most of the places where whitespace preceded '#'. I have given up and
4202 allowed the remaining 2 cases to be at the margin.
4203
4204 + Rename some variables in pcre to eliminate shadowing. This seems very
4205 pedantic, but does no harm, of course.
4206
4207 6. Moved the call to setjmp() into its own function, to get rid of warnings
4208 from gcc -Wall, and avoided calling it at all unless PCRE_EXTRA is used.
4209
4210 7. Constructs such as \d{8,} were compiling into the equivalent of
4211 \d{8}\d{0,65527} instead of \d{8}\d* which didn't make much difference to the
4212 outcome, but in this particular case used more store than had been allocated,
4213 which caused the bug to be discovered because it threw up an internal error.
4214
4215 8. The debugging code in both pcre and pcretest for outputting the compiled
4216 form of a regex was going wrong in the case of back references followed by
4217 curly-bracketed repeats.
4218
4219
4220 Version 1.02 12-Dec-97
4221 ----------------------
4222
4223 1. Typos in pcre.3 and comments in the source fixed.
4224
4225 2. Applied a contributed patch to get rid of places where it used to remove
4226 'const' from variables, and fixed some signed/unsigned and uninitialized
4227 variable warnings.
4228
4229 3. Added the "runtest" target to Makefile.
4230
4231 4. Set default compiler flag to -O2 rather than just -O.
4232
4233
4234 Version 1.01 19-Nov-97
4235 ----------------------
4236
4237 1. PCRE was failing to diagnose unlimited repeat of empty string for patterns
4238 like /([ab]*)*/, that is, for classes with more than one character in them.
4239
4240 2. Likewise, it wasn't diagnosing patterns with "once-only" subpatterns, such
4241 as /((?>a*))*/ (a PCRE_EXTRA facility).
4242
4243
4244 Version 1.00 18-Nov-97
4245 ----------------------
4246
4247 1. Added compile-time macros to support systems such as SunOS4 which don't have
4248 memmove() or strerror() but have other things that can be used instead.
4249
4250 2. Arranged that "make clean" removes the executables.
4251
4252
4253 Version 0.99 27-Oct-97
4254 ----------------------
4255
4256 1. Fixed bug in code for optimizing classes with only one character. It was
4257 initializing a 32-byte map regardless, which could cause it to run off the end
4258 of the memory it had got.
4259
4260 2. Added, conditional on PCRE_EXTRA, the proposed (?>REGEX) construction.
4261
4262
4263 Version 0.98 22-Oct-97
4264 ----------------------
4265
4266 1. Fixed bug in code for handling temporary memory usage when there are more
4267 back references than supplied space in the ovector. This could cause segfaults.
4268
4269
4270 Version 0.97 21-Oct-97
4271 ----------------------
4272
4273 1. Added the \X "cut" facility, conditional on PCRE_EXTRA.
4274
4275 2. Optimized negated single characters not to use a bit map.
4276
4277 3. Brought error texts together as macro definitions; clarified some of them;
4278 fixed one that was wrong - it said "range out of order" when it meant "invalid
4279 escape sequence".
4280
4281 4. Changed some char * arguments to const char *.
4282
4283 5. Added PCRE_NOTBOL and PCRE_NOTEOL (from POSIX).
4284
4285 6. Added the POSIX-style API wrapper in pcreposix.a and testing facilities in
4286 pcretest.
4287
4288
4289 Version 0.96 16-Oct-97
4290 ----------------------
4291
4292 1. Added a simple "pgrep" utility to the distribution.
4293
4294 2. Fixed an incompatibility with Perl: "{" is now treated as a normal character
4295 unless it appears in one of the precise forms "{ddd}", "{ddd,}", or "{ddd,ddd}"
4296 where "ddd" means "one or more decimal digits".
4297
4298 3. Fixed serious bug. If a pattern had a back reference, but the call to
4299 pcre_exec() didn't supply a large enough ovector to record the related
4300 identifying subpattern, the match always failed. PCRE now remembers the number
4301 of the largest back reference, and gets some temporary memory in which to save
4302 the offsets during matching if necessary, in order to ensure that
4303 backreferences always work.
4304
4305 4. Increased the compatibility with Perl in a number of ways:
4306
4307 (a) . no longer matches \n by default; an option PCRE_DOTALL is provided
4308 to request this handling. The option can be set at compile or exec time.
4309
4310 (b) $ matches before a terminating newline by default; an option
4311 PCRE_DOLLAR_ENDONLY is provided to override this (but not in multiline
4312 mode). The option can be set at compile or exec time.
4313
4314 (c) The handling of \ followed by a digit other than 0 is now supposed to be
4315 the same as Perl's. If the decimal number it represents is less than 10
4316 or there aren't that many previous left capturing parentheses, an octal
4317 escape is read. Inside a character class, it's always an octal escape,
4318 even if it is a single digit.
4319
4320 (d) An escaped but undefined alphabetic character is taken as a literal,
4321 unless PCRE_EXTRA is set. Currently this just reserves the remaining
4322 escapes.
4323
4324 (e) {0} is now permitted. (The previous item is removed from the compiled
4325 pattern).
4326
4327 5. Changed all the names of code files so that the basic parts are no longer
4328 than 10 characters, and abolished the teeny "globals.c" file.
4329
4330 6. Changed the handling of character classes; they are now done with a 32-byte
4331 bit map always.
4332
4333 7. Added the -d and /D options to pcretest to make it possible to look at the
4334 internals of compilation without having to recompile pcre.
4335
4336
4337 Version 0.95 23-Sep-97
4338 ----------------------
4339
4340 1. Fixed bug in pre-pass concerning escaped "normal" characters such as \x5c or
4341 \x20 at the start of a run of normal characters. These were being treated as
4342 real characters, instead of the source characters being re-checked.
4343
4344
4345 Version 0.94 18-Sep-97
4346 ----------------------
4347
4348 1. The functions are now thread-safe, with the caveat that the global variables
4349 containing pointers to malloc() and free() or alternative functions are the
4350 same for all threads.
4351
4352 2. Get pcre_study() to generate a bitmap of initial characters for non-
4353 anchored patterns when this is possible, and use it if passed to pcre_exec().
4354
4355
4356 Version 0.93 15-Sep-97
4357 ----------------------
4358
4359 1. /(b)|(:+)/ was computing an incorrect first character.
4360
4361 2. Add pcre_study() to the API and the passing of pcre_extra to pcre_exec(),
4362 but not actually doing anything yet.
4363
4364 3. Treat "-" characters in classes that cannot be part of ranges as literals,
4365 as Perl does (e.g. [-az] or [az-]).
4366
4367 4. Set the anchored flag if a branch starts with .* or .*? because that tests
4368 all possible positions.
4369
4370 5. Split up into different modules to avoid including unneeded functions in a
4371 compiled binary. However, compile and exec are still in one module. The "study"
4372 function is split off.
4373
4374 6. The character tables are now in a separate module whose source is generated
4375 by an auxiliary program - but can then be edited by hand if required. There are
4376 now no calls to isalnum(), isspace(), isdigit(), isxdigit(), tolower() or
4377 toupper() in the code.
4378
4379 7. Turn the malloc/free funtions variables into pcre_malloc and pcre_free and
4380 make them global. Abolish the function for setting them, as the caller can now
4381 set them directly.
4382
4383
4384 Version 0.92 11-Sep-97
4385 ----------------------
4386
4387 1. A repeat with a fixed maximum and a minimum of 1 for an ordinary character
4388 (e.g. /a{1,3}/) was broken (I mis-optimized it).
4389
4390 2. Caseless matching was not working in character classes if the characters in
4391 the pattern were in upper case.
4392
4393 3. Make ranges like [W-c] work in the same way as Perl for caseless matching.
4394
4395 4. Make PCRE_ANCHORED public and accept as a compile option.
4396
4397 5. Add an options word to pcre_exec() and accept PCRE_ANCHORED and
4398 PCRE_CASELESS at run time. Add escapes \A and \I to pcretest to cause it to
4399 pass them.
4400
4401 6. Give an error if bad option bits passed at compile or run time.
4402
4403 7. Add PCRE_MULTILINE at compile and exec time, and (?m) as well. Add \M to
4404 pcretest to cause it to pass that flag.
4405
4406 8. Add pcre_info(), to get the number of identifying subpatterns, the stored
4407 options, and the first character, if set.
4408
4409 9. Recognize C+ or C{n,m} where n >= 1 as providing a fixed starting character.
4410
4411
4412 Version 0.91 10-Sep-97
4413 ----------------------
4414
4415 1. PCRE was failing to diagnose unlimited repeats of subpatterns that could
4416 match the empty string as in /(a*)*/. It was looping and ultimately crashing.
4417
4418 2. PCRE was looping on encountering an indefinitely repeated back reference to
4419 a subpattern that had matched an empty string, e.g. /(a|)\1*/. It now does what
4420 Perl does - treats the match as successful.
4421
4422 ****

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

  ViewVC Help
Powered by ViewVC 1.1.5